LC Detector

This eBook is protected by Copyright law and International Treaties. All rights are reserved. This book is covered by an End UserLicensee Agreement (EULA). The full EULA may be seen at http://www.library4science.com/eula.html .

1

Chrom-Ed Book Series

Raymond P. W. Scott

LIQUID CHROMATOGRAPHY

DETECTORS


2

COPYRIGHT @2003 by LIBRARY4SCIENCE, LLCALL RIGHTS RESERVED

Neither this book or any part may be reduced or transmitted in anyform or by any means, electronic or mechanical , includingphotocopying, microfilming, and recording or by any informationstorage and retrieved system without permission in writing from thepublisher except as permitted by the in-user license agreement.

World Wide Webhttp://www.library4science.com/


3

ContentsIntroduction .............................................................................................................1

Detector Specifications.....................................................................................2Dispersion in Connecting Tubes................................................................4Low Dispersion Tubing...............................................................................7Dispersion in the Detector Sensor Volume..........................................10Apparent Dispersion from Detector Sensor Volume.......................13Dispersion Resulting from the Detector Time Constant...................16

LC Detectors Based on Refractive Index Measurement............................19The Refractive Index Detector ....................................................................20

The Angle of Deviation Method.............................................................21The Fresnel Method ..................................................................................23The Christiansen Effect Detector ...........................................................26The Interferometer Detector ...................................................................29The Thermal Lens Detector ....................................................................33The Dielectric Constant Detector...........................................................35

The UV Detectors ...............................................................................................40The UV Absorption Detectors.....................................................................40

The Fixed Wavelength UV Detector.....................................................42The Multi–Wavelength UV Detector .........................................................48

The Multi–Wavelength Dispersive UV Detector ................................49The Diode Array Detector ...........................................................................52The Fluorescence Detector...........................................................................57

The Single Wavelength Excitation Fluorescence Detector ..............59The Multi Wavelength Fluorescence Detector.........................................62

Transport Detectors............................................................................................66The Moving Wire Detector ..........................................................................67The Chain Detector........................................................................................69The Modified Moving Wire Detector.........................................................70The Disc Detector...........................................................................................76The Evaporative Light Scattering Detector.............................................78Liquid Light Scattering Detectors..............................................................81

The Low Angle Laser Light Scattering Detector..............................83The Multiple Angle Laser Light Scattering (MALLS) Detector....85

The Electrical Conductivity Detector.........................................................88The Electrochemical Detector .....................................................................93

Electrode Configurations..........................................................................94Electrode Construction.............................................................................96

The Multi–Electrode Array Detector.......................................................100References...........................................................................................................105


4


1

Introduction

Although chromatography was discovered late in the 1890s itsdevelopment was almost negligible until the 1940s and this was largelydue to the lack of an inline sensitive detector. The first, effective inlineliquid chromatography (LC) detectors were the refractive index detectorreported by Tiselius and Claesson (1) in 1942 and the conductivitydetector described by Martin and Randall (2) in 1951. These two devicesshould have evoked a growth in LC development, but, in the early fifties,gas chromatography (GC) was invented which completely eclipsed thedevelopment of LC. It was not until the early 1960s that the renaissanceof LC took place, initially based on the use of the refractive index ofTiselius and Claesson. Although a significant number of GC detectorswere developed over two or three years, the development of LCdetectors was much slower, largely due to the fact that lowconcentrations of solute in a liquid do not change the properties of aliquid nearly as much as they do a gas. In fact, the development of LCdetectors was gradual and arduous.

In a similar way to the development of GC there has been a continuousinteraction between improved detector performance and improvedcolumn performance. Initially, separations monitored by detectors withimproved sensitivity permitted a precise column theory to be developedand experimentally substantiated. This allowed new columns to bedesigned with reduced dispersion and higher efficiencies. The improvedefficiencies, however, produced small volume peaks, small, that is,compared with the volume of the detector sensor and the dispersion thattook place in the conduits of the detector system..


2

As a consequence, the ultimate efficiency obtainable from the columnwas determined by the geometry of the fluid conduits of the detectorand not its sensitivity. This provoked detector redesign, with smallersensor volumes, different geometry and shorter connecting tubesbetween the column and sensor. In turn, these modifications allowedmuch smaller particles to be used in the column resulting in even lowercolumn dispersion and higher efficiencies. In this way, just as in GC,detector design and column design have interacted over the years to apoint where the performance of LC columns are now commensuratewith those of GC columns.

Unfortunately, even today, there is no LC detector that has an equivalentperformance to the flame ionization detector (FID) used in GC. Ingeneral, LC detectors have sensitivities of two to three orders ofmagnitude less than their GC counterparts and linear dynamic rangesone to two orders of magnitude lower. Only highly specific LC detectorshave sensitivities that can approach those of GC detectors.

Detector Specifications

Detector specifications are like those for GC detectors and are listed asfollows,

1. Dynamic Range2. Response Index or Linearity3. Linear Dynamic range4. Detector Response5. Detector Noise Level6. Detector Sensitivity or Minimum Detectable Concentration7. Total System Dispersion8. Sensor Dimensions9. Detector Time Constant10. Pressure Sensitivity11. Flow Sensitivity12. Operating Temperature Range


3

In general the specifications are the same for both GC and LC detectorswith the exception of detector dispersion. Although, detector dispersionhas a minimal effect on the resolution in GC separations, detectordispersion can actually destroy a separation achieved in an LC column ifthe system is not designed correctly. Dealing with the otherspecifications, the dynamic range and linear dynamic range are the sameas those defined in book 4. The response index, the measure of detectorlinearity, can also be determined in exactly the same way, either by theincremental method of calibration, or the logarithmic dilution method. Inthe logarithmic method of calibration, mobile phase, now a liquid, ispassed continuously through an enclosed stirred vessel containing aknown mass of solute, the eluent passing directly into the detector. Thelogarithm of the detector output is plotted against the logarithm of thecalculated solute concentration and the magnitude of the response indexdetermined from the slope of the curve in the manner described in book4.

The response, noise and sensitivity are measured in exactly the same wayas for GC detectors. Pressure sensitivity and pressure tolerance have amore important significance in LC as in multidimensional LC, thedetector may be situated between two or more columns and thus musttolerate pressures up to the input pressure (e.g., several thousand p.s.i).Pressure sensitivity and flow sensitivity are also more important in LCdue to the relatively high pressures involved and the sensitivity of manysensors to pressure changes (e.g., the refractive index detector and theUV detector). However, LC columns have a high impedance to flow andso pressure pulses are often smoothed out in the column and do notreach the detector. Dispersion that takes place in a column is veryimportant and will be dealt with in some detail.

Dispersion in Detector Sensors

There are three sources of dispersion in LC detector sensors,

1. Dispersion from Connecting Tubes(Newtonian)2. Dispersion from Sensor Cell Volume (Newtonian)


4

3. Dispersion from Sensor Cell Volume ( Dilution)

Each of these sources of dispersion are controllable by careful sensordesign and employing appropriate cell geometry.

Dispersion in Connecting Tubes

The dispersion that takes place in an open tube results from the parabolicvelocity profile that occurs under conditions of Newtonian flow, (i.e.when the velocity is significantly below that which produces turbulence).Under condition of Newtonian flow, the distribution of fluid velocityacross the tube adopts a parabolic profile as shown in figure 1.

The velocity at the walls is virtually zero and that at the center amaximum. This situation is depicted diagramatically in Figure 1.

Mobile Phase Velocity

Tube Walls

Parabolic Velocity Profile

Newtonian Flow


5

Figure 1. The Parabolic Velocity Profile of a Solute Band PassingThrough a Tube

Due to the relatively high velocity at the center of the tube and the verylow velocity at the walls, the center of the band of solute passing downthe tube will move ahead of that situated at the walls. The resulting effectof band dispersion is depicted in figure 2.

Tube WallsInitial Band Width

Dispersed Band

Figure 2. Band Dispersion Resulting from Newtonian Flow

The dispersion in open tubes was examined by Golay (3) and Atwoodand Golay (4) and experimentally by Scott and Kucera (5) andLochmuller and Sumner (6). The variance per unit length of an opentube (H) according to Golay is given by

H =2Dm

u+

r2u24 Dm

where (Dm) is the diffusivity of the solute in the mobile phase,(u) is the linear velocity of the mobile phase,

and (r) is the radius of the tube.

At relatively high velocities (i.e., at velocities much greater than theoptimum velocity of the tube, which will usually be true for allconnecting tubes)

H =r 2u

24 D m


6

Furthermore, Q = p r2u

where (Q) is the flow rate through the tube.

Thus, H =Q

24 p DmNow, (H) is the variance per unit length of the tube but a more usefulparameter to the analyst is the volume variance (sv2). This can bederived using the relationship predicted by the Plate Theory (see book6).

sv2 =

V0( )2

n=

p r2 l( )2

n=

p2r 4l 2

n where (Vo) is the volume of the tube and (l) is the length of the tube

Now H =ln

, consequently sv2 = p2 r4 l H =

p r4 l Q24 D m

Thus, expression for the volume standard deviation (sv(l) ) for tubes ofdifferent length is

sv =p l Q

24 Dm

Ê

Ë Á

ˆ

¯ ˜

0.5r2 (1)

Employing equation (1) it is possible to calculate the value of (sv(l) ) fora range of cylindrical connecting tubes of different radii and differentlengths.

Table 1 Standard Deviation of Connecting Tubes of DifferentSizes

Connecting Tubes for Liquid ChromatographyStandard Deviation of Tube Dispersion

Tube Diameter l=1 cm l=2 cm l=5 cm l=10 cm l=15 cm0.001 in, 0.00254 cm 22.3 nl 31.5 nl 49.9 nl 70.5 nl 86.4 nl0.002 in, 0.00508 cm 47.6 nl 67.3 nl 106.4 nl 150 nl 184.4 nl0.003 in, 0.00762 cm 107 nl 151.3 nl 239.2 nl 0.34 ml 0.41 ml


7

0.005 in, 0.01270 cm 298 nl 421 nl 0.67 ml 0.94 ml 1.15 ml0.010 in, 0.02540 cm 1.19 ml 1.68 ml 2.66 ml 3.76 ml 4.61 ml

(Dm) is taken as 2 x 10-5 cm2sec-1 and the flow rate at 0.5 ml/min. Allvalues are fairly typical for the normal operation of the chromatographicsystem near optimum conditions.

It is seen from table 1 that the effect of dispersion in connecting tubes islarge due to the very low diffusivity of solutes in liquids. It will be shownin book 8 that for the successful use of microbore columns (columns lessthan 2 mm I.D.) tube dispersion needs to reduced to about 80 nl. Againassuming that to minimize the chance of tube blocking, the limitingminimum I.D. for the connecting tube is made to be 0.003 in (and tubesof this diameter will still easily block) then the connecting tube must beless than 1 cm long. It is clear that the length of the connecting tubebetween column and detector must be reduced to an absolute minimum.If the tubing diameter is reduced further and the column diameter isincreased, then longer tubing lengths may be possible. Alternatively, theresolution of the early peaks can be sacrificed in favor of later elutingpeaks which will also allow longer connecting tubes to be used. Thesetechniques to reduce the effect of connecting tube dispersion in LC arecommon with most manufacturers. The simple solution of designing thechromatographic system such that the detector sensor is situated veryclose to the end of the column does not appear to be considered apractical option.

Low Dispersion Tubing

In order to avoid dispersion in mobile phase conduits a number ofattempts to design low dispersion tubing has been reported. The firstattempt was by Halasz et al. (8), who crimped and bent the tube intodifferent shapes to interrupt the Newtonian flow and introduce radialflow within the tube. His devices had limited success and the tubes had atendency to block very easily.


8

In 1978 Tijssen (9), developed a theory to describe the radial flow thatwas induced into coiled tubes by the continual change in direction of thefluid as it flowed round the spirals (his theory will be considered in detailin Book 9). Tijssen found that by coiling the tubes significantly reduceddispersion, particularly at high flow rates However, the coils were a littleclumsy to form as the radius of the coil was required to be less than 3times the internal radius of the tube for optimum performance. A morepractical system was introduced by Katz and Scott (10), who developeda serpentine form of connecting tube that met the requirement that theradius of the serpentine bends (a/2 in the diagram) was less that 3 timesthat of the internal radius of the tube. A diagram of a serpentine tube isshown in figure 3.

Figure 3 Low Dispersion Tubing

During passage through the tube, the direction of mobile phase flowchanged by 180o as it passed from one serpentine bend to another.


9

1.0 2.0 3.0 4.00Flow rate (ml/min)

0.5

1.0

1.5

Var

ianc

e pe

r U

nit L

engt

h (

ml

/cm

)2

Serpentine Tube 0.10 in I.D.

Straight Tube 0.007 in !I.D.

Figure 4 Graphs of Peak Variance against Flow Rate for Straightand Serpentine TubesThis violent change in direction resulted in extensive radial flow whichaided radial transfer and greatly reduced the dispersion. This effect isclearly shown by the curves relating the variance against flow rate forstraight and serpentine tubes shown in figure 4. It is seen that at highflow rates, the dispersion is reduced by over an order of magnitude bythe serpentine tubing relative to the dispersion that occurred in thestraight tube.

Despite the apparent advantages, low dispersion serpentine tubingappears to have been employed in only one commercial LC detector. Itshould be pointed out that any conduit system that has low dispersionwill also provide very fast heat transfer rates. Serpentine tubing has beenalso used in commercial column ovens to heat the mobile phase rapidlyto the column oven temperature before it enters the column. Theserpentine tubing allows effective heat exchange with a minimum of heatexchanger volume to distort the concentration profile of the solvent


10

gradient. The different forms of dispersion profiles that are obtained fromvarious types of connecting tubes used in LC are shown in figure 4.

Dispersion Peak fromSerpentine Tube

Dispersion Peak fromCoiled Tube

Dispersion Peak fromStraight Tube(0.25mm I.D.)

Dispersion Peak fromStraight Tube(0.18mm I.D.)

Figure 4 Dispersion Profiles from Different Types of Tube

These dispersion curves were obtained using a low dispersion UVdetector (cell volume, 1.4 ml) and a sample valve with a 1 ml internal loop.All tubes were of the same length and carried the same mobile phase at aflow rate of 2 ml/min. employed. The peaks were recorded on a highspeed recorder. The peak from the serpentine tubing is seen to besymmetrical and has the smallest width. The peak from the coiled tube,although still very symmetrical is the widest at the points of inflexion ofall four peaks. The peak from the straight tube 0.25 mm I.D. is grosslyasymmetrical and has an extremely wide base width. The width andasymmetry is reduced using a tube with an I.D. of 0.18 mm but seriousasymmetry remains. Where the design of the chromatograph precludes aclose proximity between the column and the detector, the use of lowdispersion serpentine tubing may be a satisfactory alternative.

Dispersion in the Detector Sensor Volume

The finite nature of the detector sensor volume can cause peakdispersion and contribute to the peak variance by two processes. Firstly


11

there will be dispersion resulting from the Newtonian flow of fluidthrough the cell in much the same manner as the flow of a viscous fluidthrough an open tube. This will furnish a variance similar in form to thatpredicted by Golay but, as the tube length is small and the tube length toradius ratio much larger than that from a connecting tube, a differentequation is necessary to describe the dispersion effect.

Secondly, there will be a peak spreading which results from the finitevolume of the sensor. If the sensor has a significant volume, theconcentration measured will not be that entering the detector cell but theaverage concentration throughout the cell. Thus, the true profile of thepeak can not be monitored. If the sensor volume is significantly smallerthan the peak volume the effect will merely give the peak an apparentdispersion. However, if the sensor volume becomes of the same order ofmagnitude as the peak volume, then the peak profile will be distortedand resolution will be lost. In the extreme case two peaks could coexistin the sensor at one time and only a single peak will be represented.

The effect of viscous flow on dispersion will first be considered.Dispersion in Detector Sensors Resulting from Newtonian Flow

Most sensor volumes are cylindrical in shape, are relatively short inlength, and have a relatively small length-to-diameter ratio. The smalllength-to-diameter ratio is in conflict with the premises assumed in thedevelopment of the Golay equation for dispersion in an open tube.Atwood and Golay (11) extended the theory of dispersion in open tubesto tubes having small length-to-diameter ratio. The theory is complexand not relevant here as, if appropriate cell design is employed, thedispersion from viscous sources will be negligible. Nevertheless, theeffect on solute profiles is shown in figure 5.


12

0 1.0 2.0 3.0Normalized Elution Volume V/V T

Sam

ple

Con

cent

ratio

n

0.01

0.03

0.10.3

1 310

n=30

Figure 5 Elution Curves Presented as a Function of theNormalized Tube Length

It is seen that serious peak distortion can occur but as the length of thesensor cell is increased the distortion is reduced, but the dispersionincreases. However, if the conduits to the cell are appropriately designedto produce secondary flow in the cell, then the parabolic velocity profileis destroyed, and the dispersion and peak distortion eliminated. Themanner of entry of the mobile phase from the connecting conduits are,consequently, designed to produce this secondary flow and the mannerin which this is achieved is shown in figure 6.


13

Inlet from Column

Optical Window

OpticalWindow

Detector Cell

ExitMobile Phase

To Waste

Figure 6 The Design of a Modern Absorption Cell

The Newtonian flow is distorted by the manner in which the inlet andoutlet conduits are connected to and from the cell. Mobile phase entersthe cell at an angle that is directed at the cell window. It follows, that themobile phase flow has to virtually reverse its direction to pass throughthe cell producing a swirling action which introduces strong radial flowand disrupts the Newtonian flow. The effect also occurs at the exit end ofthe cell. The flow along the axis of the cell now must reverse its directionto pass out of the port which is accomplished by attaching the exitconduit at an angle to the axis of the cell. Employing this type of entryand exit connections eliminates dispersion resulting from viscous flow.

Apparent Dispersion from Detector Sensor Volume

The detector can only respond to the average value of the soluteconcentration throughout the sensor cell. At the extreme, if the sensorcell volume was large enough to contain two closely eluted peaks the


14

response would appear as a single peak, albeit very distorted in shape.This extreme condition rarely occurs, but serious peak distortion and lossof resolution can quite often happen. This will be evident when thesensor volume is of the same order of magnitude as the peak volume.The problem can be particularly severe when columns of small diameterare being used. The situation is depicted in figure 7.

0.00E+00

2.50E-01

5.00E-01

7.50E-01

1.00E+00

1.25E+00

-0.02 -0.01 0 0.01Volume Flow in ml

Cell Volume 2.5 ml

Column Length 3 cmColumn Diameter 3 mmParticle Diameter 3 mmColumn Efficiency 5000 platesk' of Solute 2

Con

cent

ratio

n in

Arb

itrar

y U

nits

Figure 7 Effect of Sensor Volume on Detector Output

Consider the elution profile of a peak eluted from a column 3 cm long, 3mm I.D. packed with particles 3 m in diameter as shown in figure 7. Ifthe peak is eluted at a (k') of 2, from figure 7 it is seen that the peakwidth at the base is about 14 ml wide. The sensor cell volume is 2.5 mland the portion of the peak in the cell is depicted in the figure. Thedetector will obviously respond to the mean concentration of the slice


15

contained in the 2.5 ml sensor volume. It is also clear that, if the sensorvolume is increased, a larger part of the peak will be contained in thecell. As a consequence, the output will be an average value of an evenlarger portion of the peak which will produce serious peak distortion.The effect of a finite sensor volume can be easily simulated with arelatively simple computer program and the output from such a programis shown in figure 8.

1.0 ml

0.1 ml

2.0 ml

3.0 ml

5.0 ml

Volume Flow of Mobile Phase

Solu

te C

once

ntra

tion

(nor

mal

ized

)

Column Length 15 cmColumn Diameter 1 mmParticle Diameter 5 mm(k') of first eluted peak 1.0

Figure 8 The Effect of Detector Sensor Volume on the Resolutionof Two Solutes

The example given, is, by far, not the worst case scenario, but is acondition where the detector sensing volume has a very serious effect onthe peak profile and, consequently, the resolution. The small bore columnproduces peaks having relatively small volumes and which arecommensurate with the volume of the sensing cell. From figure 8, it isseen that even a sensor volume as small as 1 ml will have a significanteffect on the peak width and the maximum resolution will not beobtained from the column. It is clear that the sensor cell volume must be


16

no greater than 2 ml if the column performance is not to be denigrated toa significant extent. To emphasize the effect of cell volume, it should benoted that the results from a sensor cell having a volume of 5 ml arevirtually useless. Unfortunately, many commercially available detectorshave sensor volumes as great as, if not greater than 5 ml. If small borecolumns are to be employed, such sensor volumes must be carefullyavoided.

Dispersion Resulting from the Detector Time Constant.

In addition to the sources of dispersion so far discussed, the peak canappear to be further dispersed by the combined time constant of thesensor and its associated electronics. It must be emphasized that the timeconstant of the system can not actually disperse an eluted peak, but itseffect of it on the sensor measurement can produce an apparent peakdispersion. Thus the term appear is used as the solvent profile itself is notchanged, only the profile as presented on the recorder or printer. Theeffect of the detector time constant can be calculated and the resultsfrom such a calculation are shown in figure 9.

T' = 0 sec

T' = 0.6 sec

T' = 1.5 sec

Det

ecto

r Res

pons

e

Time

Figure 9. Peak Profiles Demonstrating Distortion Resulting fromDetector Time Constant


17

The undistorted peak, monitored by a detecting system with a zero timeconstant, is about 4 seconds wide. An LC column operating at a flowrate of 1 ml/min. and having a peak base-width of 4 seconds wouldrepresent a peak with a volume of about 67 ml. It follow, that the peaksdepicted would represent those eluted fairly late in the chromatogram.However, despite the late elution, the distortion is still quite severe. Toavoid distortion of the early peaks the time constant would need to be atleast an order of magnitude less. Scott et al. (12) measured the timeconstants of two photocells and their results are shown in figure 10.

Cadmium Sulphide Photocell

IP-28 Photomultiplier

DecayCurve

DecayCurve

LogDecayCurve

LogDecayCurve

0 5 10

0 0.25 0.5Time (seconds)

Normalized Curve

Normalized Curve

LogDecayCurve


18

Figure 10. The Response Curves of Two Photocells

The output each photocell to fast transient changes in incident lightintensity was monitored with a high speed recorder. The curves for thecadmium sulfide photocell, figure 10 (chosen as an old type, sensor witha very slow response) is shown at the top of the figure. From the slopeof the log curve, the time constant was calculated to be about 2.5seconds. Such an extremely slow response would be impractical formodern chromatographic systems (i.e., two or more peaks could elutewithin the period of the time constant). The result of the slow responseof the cadmium sulfide sensor, would be to cause the peaks merge into asingle distorted peak.

The performance of the photomultiplier (representing a sensor with a fastresponse) is shown in the lower curves of figure 10. The time constant,determined from the slope of the log curve, was only 40 milliseconds. Aresponse time of 40 milliseconds is acceptable for most LC separations.Nevertheless in fast LC separations, solutes can be eluted in less than 100milliseconds in which case an even faster response might be necessary.

Contemporary sensors and electronic systems use fast solid state sensorsand solid state electronic components. Thus, most commercial detectorsystems are sufficiently fast for the vast majority of chromatographyapplications. As a general rule, the overall time constant of an LCdetecting system should be less than 50 milliseconds. For specially veryfast separations, a lower value of 15 milliseconds may be necessary. Fastsensors and electronics will respond to high frequency noise so thechromatographic system must be designed to reduce short term noise.This may involve magnetic screening to reduce the effect of stray, low-frequency electromagnetic fields from nearby power supplies and anyhigh energy consuming laboratory equipment.

In general, as the peaks in LC separations can be extremely small allsources of dispersion must be taken into account. It follows that in thedesign of the chromatograph, careful steps must be taken to minimize


19

the effect from such dispersion sources and to ensure the integrity of theseparation is maintained.

LC Detectors Based on Refractive Index Measurement

LC detectors range from those that are exclusively non specific (i.e., bulkproperty detectors, e.g., the refractive index detector) through those thatare partially specific (i.e. partial solute property detectors, e.g., the UVdetectors) to the totally specific detectors (i.e., solute property detectors,e.g., the fluorescence detector). In general, the sensitivity increasesprogressively as the detector becomes more specific, the highestsensitivities being obtained from the specific detectors.

Refractive index is a bulk property of the column eluent and so detectiondepends on the solute modifying the overall refractive index of themobile phase sufficiently to provide a signal twice that of the noise. Bulkproperty detectors have an inherently limited sensitivity irrespective ofthe instrumental technique that is used. Consider an hypothetical bulkproperty detector that monitors the density of the eluent leaving thecolumn. Assume it is required to detect the concentration of a densematerial, such as carbon tetrachloride (specific gravity 1.595), at a levelof 1 mg/ml in n-heptane (specific gravity 0.684).

Let the change in density resulting from the presence of the solute at aconcentration of 10-6 g/ml be (D d). It follows, that to a firstapproximation,

Dd =Xs d1 - d 2( )

d1

where (d1) is the density of the solute, carbon tetrachloride,(d2) is the density of the mobile phase, n-heptane

and (Xs) is the concentration of the solute to be detected.

Thus for the example given,


20

Dd =1.595 - 0.684( ) x 10-6

1.59= 5.71 x 10 -7

The coefficient of cubical expansion of n-heptane is about 1.6 x 10-3 peroC. The temperature (Dq) that would produce a change equivalent to thepresence of carbon tetrachloride at a concentration of 10-6 g/ml can,therefore, be calculated.

Thus, Dq =5.71 x 10-7

1.6 x 10 -3oC

= 3.6 x10 -4 oC

Therefore, to detect a concentration of one part per million of carbontetrachloride (at a signal to noise ratio of two), then the temperaturevariation must be maintained below 1.8 x 10-4 oC. Such temperaturestability is extremely difficult to maintain and, thus, temperature controlwill limit the sensitivity obtainable from the detector. Even the heat ofadsorption and desorption of the solute on the stationary phase canproduce temperature changes of this order of magnitude.

Similarly, the density of the contents of the cell will change with pressureand, if there is a significant pressure drop across the cell, also with flowrate. These stability problems apply to all bulk property detectors and,thus, bulk property detectors in general will all have a limited sensitivity(on average for most compounds, this will be about 10-6 g/ml). Inaddition, even to achieve this sensitivity, the sensor must always beoperated under very carefully controlled conditions.

The Refractive Index Detector

One of the first on-line detectors to be developed was the refractiveindex detector originally described by Tiselius and Claesson (14) in 1942.Despite its limited sensitivity, this detector can be very useful for


21

detecting those compounds that are nonionic, do not adsorb in the UV,and do not fluoresce.

Since 1942, there have been many types of refractive index detectorsintroduced and a number of different optical systems utilized. Only thosein common use or having particular interest will be described here.

The Angle of Deviation Method

When a monochromatic ray of light passes from one isotropic medium,(A), to another, (B), it changes its wave velocity and direction. Thechange in direction is called the refraction and the relationship betweenthe angle of incidence and the angle of refraction is given by Snell's law,namely,

n' B =n Bn A

=sin (i )sin (r)

where (i) is the angle of incident light in medium (A),(r) is the angle of refractive light in medium (B),

(nA) is the refractive index of medium (A),(nB) is the refractive index of medium (B),

and (n'B) is the refractive index of medium (B) relative to that ofmedium (A).

Refractive index is a dimensionless constant that normally decreases withincreasing temperature. The reported values are usually taken at 20o or25oC and are mean values measured for the two sodium lines. If themobile phase is allowed to flow through a hollow prism and a ray oflight passes through the prism it will be diverged from its original pathand can be focused onto a photocell. If the refractive index of the mobilephase changes, due to the presence of a solute, the angle of deviation ofthe transmitted light will also alter and the amount of light falling on thephotocell will change. A number of manufacturers have employed theangle of deviation method in refractive index detector design.


22

LightSourceMask

Lens

Sample

ReferenceAmplifier andPower Supply

ZeroAdjust

Sensor

Mirror

Recorder

Figure 11. The Refractive Index Detector Based on the Angle ofDeviation Method of Measurement

A diagram of a simple refractive index detector that is based on theangle of deviation method of measurement is shown in figure 11. Thedifferential refractometer monitors the deflection of a light beam causedby the difference in refractive index between the contents of the samplecell and those of the reference cell. A light beam from an incandescentlamp is confined to the region of the cell by an optical mask. A lenscollimates the light beam through both the sample and reference cells toa plane mirror. The mirror reflects the beam back through the sampleand reference cells to a lens which focuses it onto a photocell.

In fact, it is the location of the beam, rather than its intensity, thatchanges with the refractive index difference between the contents of thetwo cells. As the position of focus of the beam on the photoelectric cellchanges, the electrical output changes which is electronically modified toprovide a signal proportional to the concentration of solute in the cell.


23

Time

xylose

glucosesucrose

maltose

lactose

maltotriose

Figure 12. Chromatogram from an RI Detector Based on the Angleof Deviation Method of MeasurementAn example of a separation monitored by a refractive index (RI) detectorfor a typical sugar application is shown in figure 12.

The Fresnel Method

The relationship between the reflectance from an interface between twotransparent media and their respective refractive indices is given byFresnel's equation,

R =12

sin 2 (i - r)sin 2 (i + r)

+tan2 (i - r)tan2 (i + r)

È

Î Í

˘

˚ ˙

where (R) is the ratio of the intensity of the reflected light to that of theincident light and the other symbols have the meanings previouslyassigned to them. Now, sin(i)

sin(r)=

n1n 2

where (n1) is the refractive index of medium (1),and (n2) is the refractive index of medium (2).

Consequently, if medium (2) represents the column eluent, any change in(n2) will change (R) (i.e., DR) and, thus, measurement of (DR) will detect


24

changes in solute concentration. The first to utilize this principal ofdetection was in the construction of a practical detector was Conlon (4).

Conlon's device is now obsolete but it illustrates the principle of theFresnel method of detection very simply. A diagram of Conlon'sdetector is shown in figure 13.

Glass Rod

LightSource

Photocell

FromColumn

To Waste

Figure 13 A Simple Detector Based on the Fresnel Method ofRefractive Index Measurement

The sensing element consists of a rod prism sealed into a tube throughwhich the solvent flows. The rod (6.8 mm in diameter and 10 cm long)is made from a glass rod, bent to the correct optical angle (just slightlyless than the critical angle) and an optical flat is ground on the apex ofthe bend (see figure 13). The optical flat is then sealed into the windowof a flow-through cell. The photocell is arranged to be one arm of aWheatstone bridge and a reference photocell (not shown) whichmonitors light direct from the cell, is situated in another arm of thebridge.A commercial refractive index detector working on the Fresnel principleis shown diagramatically in figure 14. Light from a tungsten lamp passesthrough an IR filter (to minimize thermal effects) onto a magnifyingassembly and prism that also splits the beam into two.


25

PhotocellLamp

Focusing Lenses

FlatMirror

FlatMirror

IR Filter

Lens/SplitterFlow Cell

To UserView Port

Prism

Figure 14 A Diagram of the Optical System of a Refractive IndexDetector Operating on the Fresnel Method

The two light beams are arranged to pass through the sample andreference cells respectively. Refracted light from the mobile phase/prismsurface passes through the prism assembly and focused onto twophotocells. The prism is also arranged to reflects some light to anaperture where the surface of the prism can be observed. The photocelloutputs are electronically processed and passed to either a potentiometricrecorder or a computer data acquisition system. The refractive indexrange monitored by the device for a given prism is limited and,consequently, there are usually three different prisms available to coverthe RI ranges of 1.35–1.4, 1.31–1.44 and 1.40–1.55 respectively. Infigure 15 is shown the separation of a series of polystyrene standardsmonitored by this type of detector.


26

0 5 10 15Time (minutes)

1

2

3

4

5

67

8

Polystyrene1. MW 1,650,0002. Mw 480,0003. MW 180,0004. MW 76,0005. MW 39,0006. MW 11,6007. MW 2,9008. MW 580.

Figure 15 The Separation of Some Polystyrene Standards Using aRI Detector Operating on the Fresnel Method

The separation was carried out by size exclusion on a column packedwith 5 mm particles and operated at a flow rate 0.8 ml/min.

As a result of limited sensitivity and restricted linear dynamic range, theRI detector is only used for those applications where, for one reason oranother. all other detectors are inappropriate or impractical.This type of detector does, however, have one particular area ofapplication for which its characteristics make it particularly suitable andthat is for monitoring the separation of polymers. This is because forthose polymers containing more than six monomer units, the refractiveindex is proportional to polymer concentration and independent of itsmolecular weight. Consequently, quantitative estimation of each polymermixture can be obtained by simple normalization of peak areas and noindividual response factors are required. RI detectors have sensitivities ofabout 1 x 10-6 g/ml, a linear dynamic range of about 200 and a responseindex (r) lying between 0.97 and 1.03.

The Christiansen Effect Detector


27

This procedure for measuring refractive index arose from the work ofChristiansen on crystal filters (14,15). Consider a cell packed withparticulate material having the same refractive index as the mobile phasepassing through. If a beam of light passes through the cell there will belittle of no refraction or scattering. However, if the refractive index of themobile phase changes, there will now be a refractive index differencebetween the mobile phase and that of the packing. As a consequencesome light will be refracted away from the incident beam and theintensity of the transmitted light will be attenuated. Thus, if thetransmitted light is focused onto an appropriate photocell, then anychange in refractive index caused by the elution of a solute will producescattering and a consequent change in electrical output.

In practice, he optical dispersions of the media are likely to differ, andconsequently the refractive index will only match at one particularwavelength. As a result the fully transmitted light will be largelymonochromatic. Light of different wavelengths will be proportionallydispersed depending on the wavelength at which the two media have thesame optical dispersion. Thus, a change in mobile phase refractive willchange both the intensity of the transmitted light and its wavelength.

This device was made by GOW-MAC Inc., who claimed it had asensitivity of 1 x 10-6 refractive index units (the maximum that cold beexpected). This would be equivalent to a sensitivity of 9 x 10-6 g/ml ofbenzene (refractive index 1.501) eluted in n-heptane (refractive index1.388). The cell volume was kept to 8 ml (a little large for modernsensors) which was small enough to work satisfactorily with 4.6 mm I.D.LC columns. Different cells packed with appropriate materials werenecessary to cover the refractive index range of 1.31 to 1.60. A diagramof the Christiansen detector is shown in figure 5.


28

Reference solvent

Eluent fromColumn

ToWaste

ToWaste

SolidPacking

Prism

Achromat

Lamp

Condensor

Photocells

Aperture

Figure 16. The Christiansen Effect Detector

In the optical unit there is a pre focused lamp having an adjustablevoltage supply to allow low energy operation when the maximumsensitivity is not required. The condensing lens, aperture, achromat andbeam splitting prisms are mounted in a single tube which permitted easyoptical alignment prevented contamination from dust. The devicecontains two identical and interchangeable cells. The disadvantage of thisdetector is that the cells must be changed each time a different mobilephase is chosen in order to match the refractive index of the packing tothat of the new mobile phase. The refractive indices of the cell packingcan be closely matched to that of the mobile phase by using appropriatesolvent mixtures. In most cases solvent mixing can be achieved withoutaffecting the chromatographic resolution significantly (e.g. by replacing asmall amount of n-heptane in a mixture with either n-hexane or n-octanedepending on whether the refractive index needs to be increased ordecreased. However a considerable knowledge of the effect of differentsolvents on solute retention is necessary to accomplish this procedure


29

successfully. As a result of limitations inherent in his type of detectorcombined with the general disadvantages of the RI detector per se hasnot made the Christiansen Effect Detector very popular.

The Interferometer Detector

The interferometer detector was first developed by Bakken and Stenberg(16) in 1971. The response of the detector depends on the change in theeffective path length of a beam of light passing through a cell when therefractive index of its contents changes due to the presence of an elutedsolute. Light that has passed through the cell is focused on a photocell.Coincidentally a reference beam of light from the same source is focusedon the photocell, interference fringes are produced. The fringes changeas the path length of one light beam changes with reference to the other,thus, as the concentration of solute increases in the sensor cell during anelution of a peak, a series of electrical pulses will be generated as eachfringe passes the photocell.

The optical path length (d) of light through the cell depends on thechange in refractive index (Dn), and the path length (l), thus, d!= Dn l

In addition, the number of fringes (N) which move past a given point onthe photo cell (or the number of cyclic changes of the central portion ofthe fringe pattern) is given by,

N =2 D n l

l

where (l) is the wavelength of the light employed.The larger the value of (N) for a given (Dn), the more sensitive thedetector will be. It follows that, (l) should be made as large as possible.However, this procedure for increasing the sensitivity is limited by thedead volume of the column and the dispersion that can be toleratedbefore chromatographic resolution is impaired. A diagram of the simpleoptical system originally employed by the authors is shown in figure 17.


30

PlainMirror

HalfMirror

PlainMirror

Lamp

Photocell

FromColumn

To Waste

Sensor Cell

Figure 17 The Original Optical System Used by Bakken andStenberg in Their Interferometer Detector

Light from an appropriate source strikes a half silvered mirror and isdivided into two paths. Part of the beam is reflected by a plane mirrorback along the same path and onto a photocell.

Time


31

Figure 18 Chromatogram from the Bakken and StenbergInterferometer Detector

The other part of the beam passes through the sensor cell to a planemirror, where it is reflected back again through the sensor cell to the halfsilvered mirror that reflects it onto the photocell. Interference takes placewith the other half of the light beam on the surface of the photocell. Thetrace resulting from the elution of 8 ml of dioxane through the cell isshown in figure 18.

Each peak shown in figure 18 represents the passage of a fringe acrossthe surface of the photocell. The four interference peaks represents asingle chromatographic peak. The number of fringes will be directlyproportional to the total change in refractive index, which, in turn, willbe proportional to the total amount of solute present. In this form thedetector has limited use, but has been developed into a commerciallyviable instrument called the Optilab DSP by Wyatt Technology Inc. Adiagram of the optical system of the Optilab interference detector isshown in figure 19.

Light from the source is linearly polarized at -45o to the horizontal plane.Horizontal and vertical polarized light beams are produced and afterpassing through the Wollaston prism, one beam passes through thesample cell and the other beam through the reference cell. The samplecell beam is horizontally polarized and the reference cell beam isvertically polarized. After passing through the cells, the beams arefocused onto a second Wollaston prism and then through a quarter-waveplate which has its fast axis set -45o to the horizontal plane. A beam thatis linearly polarized in the fast axis plane after passing through the platewill lead another linearly polarized but orthogonal beam by a quarter ofa wavelength. The phase difference results in a circularly polarized beam.Each of the beams focused on the Wollaston prism consists of two suchperpendicular beams which, after the quarter wave plate, result in twocircularly polarized beams of opposite rotation. These beams willinterfere with each other to yield the original linearly polarized beam. A


32

second polarizer is placed at an angle (90 – b) to the first, allowing about35% of the signal to reach the photocell. A filter transmitting light at 546nm precedes the photocell.

LampWollaston Prism

Wollaston Prism

Sample Cell

Interference Filter

MaskPolarizer

PhotocellLens

Reference Cell

AnalyzerQuarterwave Plate

Lens

Courtesy of Wyatt Technology

Figure 19 The Optilab Interference Refractometer Detector

If the sample cell contains a higher concentration of solute than thereference cell the refractive index will be higher and the interferingbeams will be out of phase. The refractive index difference (Dn) and thephase difference (Dp) are related by

Dp =2pLDn

l

where (L) is the length of the cell,and (l) is the wavelength of the light.

The circularly polarized beams will, therefore, interfere to yield a linearlypolarized beam which is rotated

Dp2

radians, and the amplitude of the

light striking the photocell (Ap) will be given by

A p = A o cos 90 - b -D p2

Ê Ë Á ˆ

¯ ˜ = Ao cos b -

D p2

Ê Ë Á ˆ

¯ ˜


33

An extremely high sensitivity is claimed for this system but it is difficultto interpret the data in terms of minimum detectable concentration Thesmallest cell (1.4 ml) is reported to give a sensitivity of about 2 x 10-7 RIunits at a signal-to-noise ratio of two. Consequently, for benzene (RI =1.501) sensed as a solute in n-heptane (RI=1.388 ) this sensitivity wouldrepresent a minimum detectable concentration of 5.6 x 10-5 g/ml. Thealternative 7 m l cell would decrease the minimum detectableconcentration to about 1 x 10-6 g/ml, similar to that obtained for otherrefractive index detectors.

A number of LC detectors have been developed that are either based onrefractive index measurement or function on some physical property ofthe mobile phase system that is related to the refractive index. Althoughmost are not commercially available, they demonstrate the range ofsensing techniques that have been investigated as possible methods ofdetection.

The Thermal Lens Detector

If a laser is focused on an absorbing substance, the refractive index of thematerial can be modified in such a way that the medium behaves as alens. The thermal lens effect was first reported by Gorden et al. (25,26)in 1964 but since that time the phenomenon has been investigated by anumber of workers. Thermal lens formation results from extremely weaklaser light adsorption The excited-state molecules subsequently decayback to ground state causing localized temperature increases to occur inthe sample. Since the refractive index of the medium depends on thetemperature, the ensuing spatial variation of refractive index produces aneffect which appears equivalent to the formation of a lens within themedium.

For most liquids, the temperature coefficient of refractive index isnegative and consequently, the insertion of a liquid in the laser beamproduces a concave lens that results in beam divergence. Buffet andMomis (27) used the thermal lens effect to develop a small volume


34

detector, a diagram of which is shown in figure 20. The device consistsof a heating laser, the light from which is passed directly through thesample via two lens and a half mirror. Another laser, the probe laser,passes light in the opposite direction through one lens, through thesample to the half mirror where the light is reflected onto a photocell.

Heating Laser

Sample Cell

Lens Lens

HalfMirror

Probe Laser

Photocell

Pin Hole MaskHeatFilter

Filter

ReferencePhotocell

Figure 20 The Layout of a Thermal Lens Detector

A filter and a pinhole screen is placed between the mirror and the photo-cell to remove the heating laser light. When an absorbing solute is elutedfrom the column through the cell, a thermal lens is produced causing theprobe light to diverge, and the intensity of the light passing through thepinhole and on to the photocell is reduced.

The cell volume can be as little as a few microliters and, thus, would besuitable for use with microbore columns. A sensitivity of 10-6 AU hasbeen claimed for the detector and a linear dynamic range of about threeorders of magnitude. The thermal lens detector is, in fact, a special formof the refractive index detector and might, therefore, be considered auniversal detector. Nevertheless, like other bulk property detectors, it can


35

not be used with gradient elution or flow programming and hassensitivity that is no better, if as good as, other refractive indexdetectors.

The Dielectric Constant Detector

T he refractive index of a substance is a complementary property to thedielectric constant and in some circumstances is a direct function of it.For non-polar substances, the relationship between dielectric constant (e)and refractive index (n) is given by

e!!=!!n2

For semi-polar substances or mixtures of semi-polar substances and non-polar substances the Lorentz-Lorenz equation applies

e - 1e + 2

=n 2 - 1n 2 + 2

However, for polar substances and mixtures of polar and semi-polarsubstances the relationship breaks down and no simple functions describedielectric constant in terms of refractive index.

The more polar the substance, the larger is its dielectric constant. Inn o r m a l chromatography (as opposed to reversed phasechromatography) the mobile phase is normally less polar than the solutesbeing eluted. Thus, the presence of a solute in the mobile phase willincrease the dielectric constant of the mobile phase. Conversely, inreversed phase chromatography the solute is usually less polar than thesolvent and the dielectric constant of the mobile phase is reduced by thepresence of a solute. Thus. a device situated at the end of the columnwhich responds to changes in dielectric constant would act as achromatography detector. The sensor often takes the form of acylindrical or parallel plate condenser. The volume of the sensor must beas small as possible to minimize dispersion. In addition, as the sensitivity


36

of the device is proportional to the electrical capacity of the sensor, thecapacitor plates must be very close together.

A suitable circuit for use in dielectric constant measurement is anelectrical "bridge", the detector cell being situated in one arm of thebridge. If the sensor cell has a capacity greater than 100 pF, then a Weinbridge can be used; however such a cell may well have a fairly largevolume. For smaller capacity cells, the Schering bridge is moreappropriate and a diagram of a Schering bridge is shown in figure 21.

R

Ro

r

C

Cell

CoC'

D

Figure 21 The Schering Bridge for the Measurement of SmallCapacities

No capacitor is ideal, all will have some inductance and resistance inaddition to its capacity. In fact, because the plates of the capacitor aresituated in the mobile phase, if uninsulated, it is very likely to have asignificant resistance component. The current though the resistivecomponent of a conductor is in phase with the applied voltage and thecapacity component lags the applied voltage by 90o. Thus, there are twocomponents to be balanced before the output of the bridge (across (D))can be used to monitor the elution of a solute.


37

The Schering bridge is balanced by the iterative adjustment of (Ro) and(C'). At balance the following relationships will hold:

CR o

=C o

Rand r

R=

C'C o

The resistance-component of the cell reduces the bridge sensitivity tochanges in capacity and thus the plates should be well insulated toprevent conductivity through the mobile phase.

The capacity of the sensor can also be measured by making it onecomponent of a resistance/capacity or an inductance/capacity oscillator.The frequency will depend, among other things, on the capacity of thesensor and, in turn, on the dielectric constant of the material between theplates. The frequency general can be heterodyned against a referenceoscillator and the frequency difference will then be proportional to thechange in capacity and hence the dielectric constant of the mobile phase.

Poppe and Kunysten (28) described a dielectric constant detector whichincluded a reference cell for temperature compensation. The cellconsisted of two stainless steel plates 2 cm x 1 cm x 1 mm separated by agasket 50 mm thick. The two cells were identical and clamped back toback, sharing a common electrode.


38

Figure 22 The Sensor of a Dielectric Constant Detector

The device was reported to have a sensitivity of 10-6 g/ml forchloroform (e = 4.81) in n-octane. As might be expected, it was found tobe very sensitive to pressure changes in the cell (thought due to platedeformation) even when constant flow pumps were employed. The firstdielectric constant detector became commercially available in 1979 (29)and was described by Benningfield and Mowery (30). Severalapplications were reported by Bade et al. (31). A diagram of the sensoris shown in figure 22.

Each cell consisted of a concentric cylinder (inner electrode) inside alarger cylinder (the outer electrode) which formed the outer wall of thecell. Both electrodes were made of stainless steel. The two cylinders wereelectrically isolated with a cylindrical flow path through the cell. Theinner cylindrical electrodes were 1.26 cm in diameter and 0.625 cm longseparated from the outer cylinder by about 0.009 cm.


39

Time (minutes)0 5 10 15

Tripalmatin

Trumyristin

MethyleneDichloride

Figure 23 The Separation of Some Triglycerides Monitored by aDielectric Constant Detector

The linear dynamic range of the detector was reported to be 3.5 x 104 .The sensitivity was quoted as about 1 x 10-7 g/ml, which would be closeto the theoretical limit for bulk property detectors. An example of theuse of the dielectric constant detector to monitor a separation oftriglycerides is shown in figure 23.

Bulk property detectors have neither the sensitivity nor the lineardynamic range of solute property detectors and are less frequently usedin modern LC analyses. None can be used satisfactorily with gradientelution, flow programming or temperature programming and so theyrestrict the choice of development. They do have certain unique areas of


40

application, some of which have already been mentioned. Their useprobably represents less than 5% of all LC analyses.

The UV Detectors

Although over the years a large number of LC detectors have beendeveloped and described, the vast majority of all contemporary LCanalyses are carried out using one of four detectors, the UV detector inone of its forms, the electrical conductivity detector, the fluorescencedetector and the refractive index detector. In addition, some form of theUV detector probably accounts for 80% of those analyses.

The UV Absorption Detectors

UV absorption detectors respond to those substances that absorb light inthe range 180 to 350 nm. Many (but not all) substances absorb light inthis wavelength range, including those substances having one or moredouble bonds (p electrons) and substances having unshared (unbonded)electrons, e.g. all olefins, all aromatics and compounds, for example,containing > C = O , > C = S , – N = N – groups. The sensor of a UVdetector consists of a short cylindrical cell having a capacity between 1m l and 10 m l through which passes the column eluent. UV light isarranged to pass through the cell and fall on a photo–electric cell (orarray). The output from the photocell passes to a modifying amplifierand then to a recorder or data acquisition system.

The relationship between the intensity of UV light transmitted through acell (IT) and the concentration of solute contained by it (c) is given byBeer's Law. IT = Ioe -klc

or ln (IT) = ln (Io) - kcl

where (Io) is the intensity of the light entering the cell,(l) is the path length of the cell,


41

and (k) is the molar extinction coefficient of the solute for thespecific wavelength of the UV light.

Differentiating,

d log IT

Io

Ê

Ë Á

ˆ

¯ ˜

d c = - k l

The sensitivity of the detector, as measured by the transmitted light, willbe directly proportional to the extinction coefficient (k) and the pathlength of the cell (l). To increase the sensitivity of the system, (l) must beextended but there is a limit to which (l) can be increased as the cellvolume and, in particular, the length of the cell must be restricted. This isnecessary to minimize peak dispersion in the sensor and to avoid morethan a small fraction of a peak existing in the cell at any one time Thisproblem has already been discussed. To restrict peak dispersion, theradius of the cell must also be reduced as (l) is increased. Thus, less lightwill fall on the photo–cell, the signal–to–noise ratio will be reduced andthus the detector sensitivity or minimum detectable concentrationdenigrated. Thus, increasing the detector sensitivity by increasing thepath length has limitations and a well–designed cell involves a carefulcompromise between cell radius and length to provide the maximumsensitivity. Most modern UV detector sensors have path lengths thatrange between 1 and 10 mm and internal radii that range from about 0.5to 2 mm

Now, LogITIo

= k' l c = A

where (A) is termed the absorbenceNow (DA) is sometimes employed to define the detector sensitivitywhere the value of (DA) is the change in absorbence that provides asignal-to-noise ratio of two.

Thus D A = k' l D c


42

where (D c) is the detector concentration sensitivity or minimumdetectable concentration.

Thus D c =DAk' l

Thus. two detectors, having the same sensitivity defined as the minimumdetectable change in absorbence, will not necessarily have the samesensitivity with respect to solute concentration. Only if the path lengthsof the two sensors are identical will they also exhibit the sameconcentration sensitivity. This can cause some confusion as it would beexpected that two instruments having the same spectroscopic sensitivitywould also have the same chromatographic sensitivity. To compare thesensitivity of two detectors given in units of absorbence the path lengthsof the cells in each instrument must be taken into account.

UV detectors can be used with gradient elution providing the solvents donot absorb significantly over the wavelength range that is being used fordetection. In reversed phase chromatography, the solvents usuallyemployed are water, methanol, acetonitrile and tetrahydrofuran (THF), allof which are transparent to UV light over the total wavelength rangenormally used by UV detectors. In normal phase operation more care isnecessary in solvent selection as many solvents that might be appropriateas the chromatographic phase system are likely to absorb UV light verystrongly. The n-paraffins, methylene dichloride, aliphatic alcohols andTHF are useful solvents that are transparent in the UV and can be usedwith normal distribution systems (e.g. a polar stationary phase such assilica gel).

The Fixed Wavelength UV Detector

The fixed wavelength UV detector uses light of a single wavelength (ornearly so) which is produced by a specific type of discharge lamp.


43

200 250 300 350Wavelength in nm

100

10

10

100

100

10

Rel

ativ

e Em

issio

n

Mercury Lamp

253.7

302.2 313.2

ZincLamp

213.9

277.1280.1

307.6

214.4226.5

228.8

283.6

288.1293.1 308.1

313.2

325.4

326.1340.3346.6

Cadmium Lamp

Figure 24 Emission Spectra for Three Discharge Lamps

The most popular lamp is the low pressure mercury vapor lamp, whichgenerates most of its light at a wavelength of 254 nm. Other lamps thatcould be used are the low-pressure cadmium lamp which generates themajority of its light at 225 nm and the low pressure zinc lamp that emitslargely at 214 nm. None of the lamps are strictly monochromatic and


44

light of other wavelengths is always present but usually at a significantlylower intensity. The emission spectra of the mercury, cadmium and zinclamps are shown in figure 24. It is seen that to obtain monochromaticlight an appropriate filter would be needed. The low pressure mercurylight source (wavelength 253.7 nm) provides the closest to truemonochromatic light of all three lamps. However, there is light presentof significant intensity below 200 nm, but light of such wavelengths isgenerally absorbed by the mobile phase.

The zinc lamp has a major emission line at 213.9 but the emission line at307.6 is of comparable intensity and a suitable filter would be needed ifdetection was required to be exclusively at the lower wavelength. Thecadmium lamp has a major emission line at 228.8 but light is emitted atboth lower wavelengths and at substantially higher wavelengths and soan appropriate filter would again be desirable. Suitable interference filterscan be quite expensive to construct, which may account for theunpopularity of these two lamps. They do, however, emit light atwavelengths which would give an increased sensitivity to substancessuch as proteins and peptides, which might make their use worthwhile inthe biotechnology field. A diagram of a typical optical system for a fixedwavelength UV detector is shown in figure 25. Light from the UVsource is collimated by a suitable lens and passed through both thesample cell and the reference cell and then on to two photo cells Thecells are cylindrical with quartz windows at either end. The reference cellcompensates for any absorption that mobile phase might have at thesensing wavelength. The outputs from the two photo cells are passed toa signal modifying amplifier so that output is linearly related to theconcentration of solute being detected. For reasons already discussed,modern sensor cells have angular conduits that form a 'Z' shape toreduce dispersion. The UV sensor can be sensitive to both flow rate andpressure changes but this instability can be greatly reduced if the sensoris well thermostatted.


45

Low PressureMercury Lamp

Sample cell

Reference Cell

FromColumn

To Waste

Quartz Lens

QuartzWindows

PhotoCells

QuartzWindows

To Waste

Reference Flow

Figure 25. The Fixed Wavelength UV Detector

The fixed wavelength UV detector is one of the most commonly usedLC detectors; it is sensitive, linear and relatively inexpensive. Sensitivity(minimum detectable concentration) can be expected to be about 5 x10–8 g/ml with a linear dynamic range of about three orders ofmagnitude for 0.98 < r < 1.02. The separation of some aromatichydrocarbons by exclusion chromatography on a very high efficiencycolumn (efficiency ca 250,000 theoretical plates) monitored by a fixedwavelength detector is shown in figure 26. All the solutes are distinctlyresolved despite their having molecular weight differences equivalent toonly two methylene groups. The peaks from such columns are only afew microliters in width and so a specially reduced volume sensor cellwas necessary to cope with the high efficiencies and allow theconsequent improved resolution to be realized. The molecular weight of


46

decyl benzene is 218 and, thus, one methylene group would represent adifferential of only 6.4% of the molecular weight.

Column, length, 10m, I.D., 1 mm, mobile phase tetrahydrofuran, flow-rate 30 ml/min.,adsorbent Partisil 10. Column efficiency ca. 250,000 theoretical plates. Solutesbenzene, ethyl benzene, butyl benzene, hexyl benzene, octyl benzene and decylbenzene. Such columns must be packed in short lengths (about 1 m) andsubsequently joined and are thus somewhat tedious to construct.

Figure 26. The Separation of Some Alkyl Benzenes by Exclusionon a High Efficiency Column.

A very simple fixed wavelength detector suitable for use in preparativechromatography is shown in figure 27. This detector was invented byMiller and Strusz (32) and originally manufactured by GOW-MACInstruments. As opposed to detectors used for analytical purposes,detectors for preparative work need to have a very low sensitivity assample sizes are large and consequently the solute concentrations arevery high. If analytical detectors are used for preparative work a portionof the eluent is split from the main stream, diluted with more mobilephase and then passed through the detector. In practice, this is a ratherawkward procedure. As seen in figure 27 the column eluent passesthrough a delivery tube and onto a supporting plate that is usually made


47

of fused quartz, so that adequate UV light can reach the photo cellplaced on the other side of the plate.

FromColumn

FromColumn

GlassPlate

GlassPlate

UV Lamp

PhotoCellMaskFlow of Mobile Phase

Over Plate

Figure 27. Fixed Wavelength Detector for Preparative Work

The liquid flows over the plate and the effective path length of the sensorwill be the film thickness which will be unique to the particular solventused as the mobile phase. The UV lamp is situated above the upper sideof the plate and the photo cell on the lower side. A reference photo cell(not shown) is situated close to the lamp and the output used tocompensate for changes in light intensity from variations in lampemission. The short path length results in a low sensitivity and thedetector can operate satisfactorily at concentrations as high as 10-2 g/ml(1% w/w), which is ideal for preparative chromatography. A particularadvantage of this type of sensor is its very low flow impedance and thus


48

can easily accommodate the high flow rates used in preparative LC. Thefilm thickness does depend, among other factors, on the column flowrate, consequently, precise flow control is necessary for the detector toperform satisfactorily.

The Multi–Wavelength UV Detector

The multi–wavelength detector employs a light source that emits lightover a wide range of wavelengths. Employing an appropriate opticalsystem (a prism or diffraction grating), light of a specific wavelength canbe selected for detection purposes. Light of a specific wave lengthwavelength might be chosen where a solute has a absorption maximumto provide maximum sensitivity. Alternatively, the absorption spectra ofan eluted substances could be obtained for identification purposes byscanning over a range of wavelengths. The latter procedure, however,differs with the type of multi–wavelength detector being used.

There are two basic types of multi–wavelength detector, the dispersiondetector and the diode array detector, the latter being the more popular.In fact, very few dispersion instruments are sold today but many are stillused in the field and so their characteristics will be discussed. Allmulti–wavelength detectors require a broad emission light source such asdeuterium or the xenon lamp, the deuterium lamp being the mostpopular.

The two types of multi-wavelength detectors have important differences.In the dispersive instrument, the light is dispersed before it enters thesensor cell and thus virtually monochromatic light passes through thecell. However, if the incident light is of a wavelength that can excite thesolute and cause fluorescence at another wavelength, then the lightfalling on the photo cell will contain the incident light together with anyfluorescent light that may have been generated. It follows, that the lightmonitored by the photo cell may not be monochromatic and light ofanother wavelength, if present, would impair the linear nature of theresponse. This effect would be negligible in most cases but with certainfluorescent materials the effect could be significant. The diode arraydetector operates quite a differently. Light of all wavelengths generated


49

by the deuterium lamp is passed through the cell and then dispersed overan array of diodes. Thus, the absorption at discrete groups of wavelengthsis continuously monitored at each diode. However, light falling on adiscrete diode may not be derived solely from the incident light but maycontain light generated by fluorescence excited by light of a shorterwavelength. Unfortunately, this effect is exacerbated by the fact that thecell contents are exposed to light of all wavelengths emitted by thesource and so fluorescence is more likely. Thus, under somecircumstances, measurement of transmitted light may involve fluorescentlight and the absorption spectrum obtained for a substance may be adegraded form of the true absorption curve.

The ideal multi–wavelength detector would be a combination of both thedispersion system and the diode array detector. This arrangement wouldallow a true monochromatic light beam to pass through the detector andthen the transmitted beam would itself be dispersed again onto a diodearray. Only that diode sensing the wavelength of the incident light wouldbe used for monitoring the transmission. In this way any fluorescent lightwould strike other diodes, the true absorption would be measured andaccurate monochromatic sensing could be obtained.

The Multi–Wavelength Dispersive UV Detector

A diagram of the multi–wavelength dispersive UV detector is shown infigure 28. Light from the deuterium lamp is collimated by two curvedmirrors onto a holographic diffraction grating. The dispersed light is thenfocused by means of a curved mirror, onto a plane mirror and light of aspecific wavelength is selected by appropriately positioning the angle ofthe plane mirror. Light of the selected wavelength is then focused bymeans of a lens through the flow cell and, consequently, through thecolumn eluent. The exit beam from the cell is then focused by anotherlens onto a photo cell which gives a response that is some function of theintensity of the transmitted light. The detector is usually fitted with ascanning facility that allows the spectrum of the solute contained in thecell to be obtained. There is an inherent similarity between UV spectra of


50

widely different types of compounds, and so UV spectra are not veryreliable for the identification of most solutes.

CurvedMirror

QuartzLenses

CurvedMirror

Deuterium Lamp

Diffraction Grating

PlaneMirror

PlaneMirror

FromColumn

To Waste

Photo Cell

Sample Cell

Courtesy of the Perkin Elmer Corporation

Figure 28 The Multi–Wavelength Dispersive UV Detector

The technique can be used, however, to determine the homogeneity of apeak (e.g., by comparing spectra taken from both sides of the peak.Both spectra are normalized and either one is subtract one from theother and the difference is shown to be zero, or the ratio of the twospectra is calculated and the result shown to be unity.

A common use of multi-wavelength choice is to enhance the sensitivityof the detector by selecting a wavelength that is characteristicallyabsorbed by the substance of interest. Conversely, a wavelength can bechosen that substances of little interest in the mixture do not adsorb and,thus, make the detector more specific to those substances that do. Anexample of the use of the variable wavelength UV in this way is afforded


51

by the separation of some carboxylic acids that is monitored by UVabsorption at 210 nm. The separation is shown in figure 29. Theseparation of a series of common fatty acids was carried out on areversed phase column using water buffered with phosphoric acid as themobile phase.

Time (minutes)

12

3

45 6

7 8

Column: Spherisorb® Octyl, 25 cm x 4.6 mm I.D., 5 mm particles. Mobile Phase: 0.2M phosphoric acid. Flow rate 0.8 ml/min. monitored at 210 nm. 1. tartaric acid, 2.lactic acid, 3. malic acid, 4. formic acid, 5. acetic acid, 6. citric acid, 7. succinic acid, 8.fumaric acid.

Courtesy of Supelco Inc.

Figure 29. The Separation of Some Carboxylic Acids Monitored byUV Absorption at 210 nm

Multi-wavelength dispersive detectors has proved extremely useful,providing adequate sensitivity, versatility and a linear response. As aresult of the need for an optical bench inside the instrument, however, itis somewhat bulky In addition, it has mechanically operated wavelengthselection and requires a stop/flow procedure to obtain spectra "on-the-fly". In contrast, the diode array detector has the same advantages butnone of the disadvantages.


52

The Diode Array Detector

The diode array detector also utilizes a deuterium or xenon lamp thatemits light over the UV spectrum range. Light from the lamp is focusedby means of an achromatic lens through the sample cell and onto aholographic grating. The dispersed light from the grating is arranged tofall on a linear diode array.

The resolution of the detector (Dl) will depend on the number of diodes(n) in the array, and also on the range of wavelengths covered (l2 - l1).

Thus Dl =l2 - l1

n

Consequently, the ultimate resolving power of the diode array detectorwill depend on the semi–conductor manufacturer and on how narrowthe individual photo cells can be commercially fabricated.

A diagram of a diode array detector is shown in figure 30.


53

Deuterium Lamp

AchromaticLens System

Shutter

Outlet

Eluent fromColumn

FlowCell

Holographic Grating

Photo Diode Array

Figure 30. The Diode Array Detector

Light from the broad emission source is collimated by an achromaticlens system so that the total light passes through the detector cell onto aholographic grating. In this way the sample is subjected to light of allwavelengths generated by the lamp. The dispersed light from the gratingis allowed to fall onto a diode array. The array may contain manyhundreds of diodes and the output from each diode is regularly sampledby a computer and stored on a hard disc. At the end of the run, theoutput from any diode can be selected and a chromatogram producedusing the UV wavelength that was falling on that particular diode.During chromatographic development, the output of one diode isrecorded in real time producing a real time chromatogram. It is seen thatby noting the time of a particular peak, a spectrum can be obtained byrecalling from memory the output of all the diodes at that particulartime.


54

Ratio = 2.5

I max = 274 nm P.I. =0

Absorbance 254 nm

Ratio A \A 235 245

The chlorthalidone was isolated from a sample of tablets and separated by a reversephase (C18) on a column 4.6 mm I.D., 3.3 cm long, using a solvent mixtureconsisting of 35% methanol, 65% aqueous acetic acid solution (water containing 1%of acetic acid). The flow rate was 2 ml/min.

Courtesy of the Perkin Elmer CorporationFigure 31. Dual Channel Plot from a Diode Array DetectorConfirming Peak Purity

The diode array detector can be used in a number of unique ways andan example of the use of a diode array detector to verify the purity of agiven solute is shown in figure 31. The chromatogram monitored at 274nm is shown in the lower part of figure 31.As a diode array detector wasemployed, it was possible to ratio the output from the detector atdifferent wavelengths and plot the ratio simultaneously with thechromatogram monitored at 274 nm. If the peak was pure, the ratio ofthe adsorption at the two wavelengths (those selected were 225 and 245nm) would remain constant throughout the elution of the entire peak.The upper diagram in figure 31 shows this ratio plotted on the same timescale and it is seen that a clean rectangular peak is observed whichunambiguously confirms the purity of the peak for chlorthalidone. The


55

wavelength chosen to provide the confirming ratio will depend on theUV adsorption characteristics of the substance concerned, relative tothose of the most likely impurities to be present, consequently thewavelengths must be chosen with some circumspection.

Anthracene

The separation was carried out on a column 3 cm long, 4.6 mm in diameter andpacked with a C18 reversed phase on particles 3 m in diameter.


Figure 32. The Separation of Some Aromatic Hydrocarbons

Another interesting example of the use of the diode array detector toconfirm the integrity of an eluted peak is afforded by the separation ofthe series of hydrocarbons shown in figure 32.The separation appears tobe satisfactory and all the peaks appear to represent individual solutes;without further evidence, it would be reasonable to assume that all the

peaks were pure. However, by plotting the adsorption ratio, 250 nm255 nm

, for

the anthracene peak it becomes apparent that the peak tail contains animpurity as the clean rectangular shape of the peak top is not shown. Theabsorption ratio peaks are shown in figure 33.


56

The ratio peaks depicted in figure 33 clearly indicate the presence of animpurity by the sloping top of the anthracene peak. This is furtherconfirmed by the difference in the spectra for the leading and trailingedges of the peak.

250/

255

nm R

atio

Irregular top to peak showing presence of impurity

Figure 33 Curves Relating the Adsorption Ratio, 250 nm255 nm

, and

TimeSpectra taken at the leading and trailing edge of the anthracene peak areshown superimposed in figure 34. Further work identified the impurityas t-butyl benzene at a level of about 5%.

Spectra of Trailing Edge of Peak

Spectra of Leading Edge of Peak

Ads

orpt

ion

Uni

ts


57

Figure 34 Superimposed Spectra Taken at the Leading andTrailing Edges of the Anthracene Peak

Another example of the use of the diode array detector to demonstratepeak purity is shown in figure 35.

Time (minutes)

a b

cd

e

200 240 280 320(nm)

190 230 270 310(nm)

mAUImpure

amAU

b

Pure

Figure 35 Diode Array Spectra Demonstrating Peak Purity

A chromatogram is shown containing five peaks and spectra have beentaken of peak (a) and peak (b) halfway up the rising side of each peak, atthe top of each peak and halfway down the trailing edge of each peak.The spectra are also included in the figure 35. The impure and pure peakare unambiguously identified illustrating the value of this type ofdetecting system for analytical purposes.

The performance of both types of multi-wavelength detectors are verysimilar and typical sensitivities would be about 1 x 10-7g/ml (significantlyless than the fixed wavelength detector) with a linear dynamic range ofabout 5000 and a response index lying between 0.97 and 1.03.

The Fluorescence Detector


58

When light is emitted by molecules that are excited by electromagneticradiation, the phenomenon is termed photoluminescence. If the release ofelectro-magnetic energy is immediate, or stops on the removal of theexcitation radiation, the substance is said to be fluorescent. If, however,the release of energy is delayed, or persists after the removal of theexciting radiation, then the substance is said to be phosphorescent.Fluorescence has been shown to be extremely useful as a detectionprocess and detectors based on fluorescent measurement have providedsome of the highest sensitivities available in LC.

When a molecule adsorbs light, a transition to a higher electronic statetakes place and this absorption is highly specific for the moleculesconcerned; radiation of a specific wavelength or energy is only absorbedby a particular molecular structure. If electrons are raised to an upperexcited single state, due to absorption of light energy, and the excessenergy is not immediately dissipated by collision with other molecules orby other means, light will be emitted at a lower frequency as the electronreturns to its ground state and the substance is said to fluoresce. As someenergy is always lost before emission occurs then, in contrast to Ramanscattering, the wavelength of the fluorescent light is always greater thanthe incident light.

Detection techniques based on fluorescence affords greater sensitivity tosample concentration, but less sensitivity to instrument instability, (e.g.sensor temperature and pressure). This is due to the fluorescent lightbeing measured against a very low light background (i.e., against a verylow noise level). This is opposite to light absorption measurements wherethe signal is superimposed on a strong background signal carrying a highnoise level. Unfortunately, relatively few compounds fluoresce in apractical range of wavelengths. However, some compounds, includingproducts from foods, drugs, dye intermediates etc., do exhibitfluorescence and can be monitored by fluorescent means. In addition,many substances can be made to fluoresce by forming appropriatederivatives.


59

The optical system of most fluorescent detectors is arranged such thatthe fluorescent light is viewed at an angle to the exciting incident lightbeam. This minimizes the amount of incident light that can interfere withthe fluorescent signal. Under such circumstances, the fluorescent signal isviewed against a an almost black background and thus, furnishes themaximum signal to noise. A filter can be used to reduce the backgroundlight still further by the removal of any stray scattered incident light.

The fluorescence signal (If) is given by

I f = f Io 1 – e -k c l( )

where (f) is the quantum yields (the ratio of the number of photonsemitted and the number of photons absorbed),

(Io) is the intensity of the incident light,(c) is the concentration of the solute,(k) is the molar absorbence,(l) is the path length of the cell.

Fluorescence detectors can be simple or complex, the simplest consists ofa single wavelength excitation source and a sensor that monitorsfluorescent light of all wavelengths. For certain samples, this form offluorescence detector can be very sensitive and relatively inexpensive.However, employing excitation light of a single wavelength and only abroad emission wavelength, it is not very versatile. Conversely, thefluorescence spectrometer fitted with a small sensor cell is far morecomplex but with both selectable excitation wavelengths and emissionwavelengths is extremely versatile. In addition, excitation and emissionspectra can be obtained as required.

The Single Wavelength Excitation Fluorescence Detector


60

The single wavelength excitation fluorescence detector is probably themost sensitive LC detector that is available, but is achieved by forfeitingversatility. A diagram of a simple form of the fluorescence detector isshown in figure 36. The excitation light is normally provided by a lowpressure mercury lamp which is comparatively inexpensive and providesrelatively high intensity UV light at 253.7 nm. Many substances thatfluoresce will be excited by light of this wavelength.

QuartzWindow

QuartzWindow

Photo Cell

Lens

FromColumn

Fluorescent Light To Waste

UV Light Source

Excitation Light

Figure 36. The Single Wavelength Excitation Fluorescent Detector

The excitation light is focused by a quartz lens through the cell. Asecond lens, set normal to the incident light, focuses the fluorescent lightonto a photo cell. A fixed wavelength fluorescence detector will have asensitivity (minimum detectable concentration at an excitationwavelength of 254 nm) of about 1 x 10-9 g/ml and a linear dynamicrange of about 500 with a response index of 0.96 < r <1.04.


61

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

Column: 2 Pecosphere™–5C C18 (150 mm x 4.6 mm) in series. Mobile Phase: 90%acetonitrile/10% water. Flow rate: 2.0 ml/min. Detector Fluorescence (Excitation 254nm total emission sensed). Sample: 20 ml of NBS Standard.1. Naphthalene 2. Fluorene 3. Acenaphthene4. Phenanthrene 5. Anthracene 6. Fluoranthracene7. Pyrene 8. Benzo(a)anthracene 9. Chrysene10. Benzo(b)fluoranthene 11. Benzo(k)fluoranthene 12. Benzo(k)fluoranthene13. Dibenz(a,h)anthracene 14.Indeno(1,2,3,cd)pyrene 15. Benzo(ghi)perylene

Courtesy of the Perkin Elmer CorporationFigure 37. Separation of the Priority Pollutants Monitored by theSimple Fluorescence Detector


62

An example of a separation monitored by a simple fluorescence detectoris the separation of a mixture of priority pollutants shown in figure 37.The excitation light was approximately monochromatic at 254 nm andall wavelengths of the fluorescent light was sensed by the photo cell.There are some compromises between the expensive fluorescencespectrometer detector and the single wavelength excitation fluorescencedetector. Some have a single monochromators that select the wavelengthof the excitation light, others employ a single monochromator to selectthe emission wavelength or provide emission spectra.

The Multi Wavelength Fluorescence Detector

The multi wavelength fluorescence detector contains twomonochromators, one to select the excitation wavelength and the secondto select the fluorescence wavelength or produce a fluorescencespectrum A diagram of the multi wavelength fluorescence detector isshown in figure 38.

The detector comprises a fluorescent spectrometer fitted with suitableabsorption cell that is sufficiently small so as not to degrade theresolution of an LC column. There are two distinctly different light pathsone for the excitation light and one for the emitted light. The differentlight paths are depicted separately, the excitation light in dark blue andthe emitted light in light blue.

The broad band excitation source (usually a deuterium lamp) is placed atthe focal point of an ellipsoidal mirror (shown at the top left hand cornerof the diagram). The resulting parallel light beam falls on a toroidalmirror that focuses it onto the grating on the left-hand side of thediagram. This grating selects the wavelength of the excitation light. Lightof the selected wavelength passes to a spherical mirror and then to aellipsoidal mirror (shown at the base of the diagram) which focuses itonto the sample. The excitation light path is in dark blue and is situatedon the left-hand side of the diagram. A beam splitter is situated betweenthe spherical mirror and the ellipsoidal mirror (in the center of the


63

diagram) which reflects a portion of the incident light onto anothertoroidal mirror which focuses it onto the reference photo cell.

ExcitationLight SourceEllipsoidal

Mirror

Toroidal Mirror Photo

Cell Spherical Mirror

Toroidal Mirror

Spherical Mirror

Grating that Selects the Wavelength of the Excitation Light

BeamSplitter

Ellipsoidal Mirrors

Sample Cell

Grating that Selects the Wavelength of the Emitted Light

Reference Cell

Figure 38. The Fluorescence Spectrometer Detector

The path of the fluorescent light is in light blue and is largely on the righthand side of the diagram. Fluorescent light from the cell is focused by anellipsoidal mirror on to a spherical mirror which then focuses the lightonto a grating situated (seen at about center right of the figure). Thegrating selects the specific wavelength of the fluorescent light to bemonitored. Light of the selected wavelength passes to a photoelectric cellwhich monitors its intensity.


64

The instrument is quite complex but is extremely versatile. The use ofthe detector to optimize both the excitation light and the fluorescencelight to provide high selectivity for the Fluoropa derivative of neomycinis shown in figure 39. It is a very good example of the selection of aspecific excitation light wavelength and the complementary emissionlight wavelength to provide maximum sensitivity.

Neomycin

0 2 4 6Time (minutes)

Column: Supercosil LC–8, 15 cm x 4.6 mm, 5 mm particles: Mobile Phase:tetrahydrofuran : 0.0056M sodium sulfate/0.007M acetic acid/0.01M pentanesulfonate, 3:97. Flow rate: 1.75 ml/min. Post Column reagent: 1L 0.4M boricacid/0.38M potassium hydroxide containing 6 ml 40% Brij–35, 4 ml mercaptoethanol,0.8g o-phthalaldehyde. Flow rate 0.4 ml/min. Mixer 5 cm x 4.6 mm column packedwith glass beads. Reactor 10 ft x 0.5 mm knitted Teflon capillary tubing. ReactionTemperature 40oC. Sample 20 ml of a mobile phase extract of a commercial sample.Excitation wavelength 365 nm;emission wavelength 418 nm.

Courtesy of Supelco Inc.

Figure 39. Detection of Neomycin OPA Derivative at an ExcitationWavelength of 365 nm and an Emission Wavelength of 418 nm

Optimizing excitation and emission light wavelengths to obtainmaximum sensitivity for a complex mixture can become quite involvedas shown by the separation of some priority pollutants depicted in figure


65

40. The separation was carried out on a column 25 cm long, 4.6 mm indiameter and packed with a C18 reversed phase.

Time in Seconds

12

3

4

5

6 7

89 10 11

12 13

14

15

Fluo

resc

ence

Inte

nsity

1 Naphthalene 9 Chrysene2 Acenaphthene 10 Benzo(b)fluoranthene3 Fluorene 11 Benzo(k)fluoranthene4 Phenanthrene 12 Benzo(a)pyrene5 Anthracene 13 Dibenz (a,h)anthracene6 Fluoranthene 14 Benzo(ghi)perylene7 Pyrene 15 Indeno(123-cd)pyrene8 Benz(a)anthracene

Fluorescence Detector Program Time (seconds)

Wavelength of Excitation Light

Wavelength of Emitted Light

0 280 nm 340 nm 220 290 nm 320 nm 340 250 nm 385 nm 510 260 nm 420 nm 720 265 nm 380 nm 1050 290 nm 430 nm 1620 300 nm 500 nm


66

Figure 40. Separation of a Series of Priority Pollutants withProgrammed Fluorescence DetectionThe mobile phase was programmed from a 93% acetonitrile, 7% waterto 99% acetonitrile, 1% water over a period of 30 minutes. The gradientwas linear and the flow rate was 1.3 ml/min. The separation illustrates theclever use of wavelength programming to obtain the maximumsensitivity. During development both the wavelength of the excitationlight and that of the emission light were changed to provide maximumsensitivity for the particular solute.

The detector can provide fluorescence or excitation spectra by arrestingthe flow of mobile phase when the solute resides in the detecting cell andscanning either the excitation or fluorescent light. (This is the sametechnique as that used to provide UV spectra with the variablewavelength UV detector). As a consequence, it is possible to obtainexcitation spectra at any chosen fluorescent wavelength or fluorescentspectra at any chosen excitation wavelength. Thus, even with relativelypoor spectroscopic resolution many hundreds of spectra can beproduced, any or all of which (despite many spectra being very similar)can be used to confirm the identify a compound.

Transport Detectors

A transport detector consists of a carrier such as a metal chain, wire ordisc that passes continually through the column eluent extracting asample of the mobile phase containing the solute as a thin film adheringto its surface. The mobile phase is eliminated by evaporation leaving thesolute as a coating on the carrier. The carrier is then scanned by anappropriate sensing technique to monitor the residual solute. Forexample, an FID could be used to sense the pyrolysis products of thesolute by heating the carrier and most of the pyrolysis productscontaining carbon would be detected. The system is obviously restrictedto those solutes that are involatile and, in addition, the solvents used forthe mobile phase must be volatile (and extremely pure). The formercondition is usually met in LC, otherwise the analysis would probably be


67

carried out by GC. The latter condition is normally easy to achieve asthere is a wide choice of solvents readily available for LC.

The system appears to be ideal but the early models had somedisadvantages. The instruments were bulky, expensive and someincorporated a 90strontium source (for the argon detector) all of whichwere unpopular. In addition, as a result of the basic design, theanticipated high sensitivity was not realized and the apparatus wasclumsy and difficult to operate. However, as it was a universal detectorand was unaffected by the solvents used, it was readily accepted by thesoap and cosmetic industry.

The Moving Wire Detector

The original wire transport detector was developed by James et al. (33)and subsequently manufactured and marketed by Pye Unicam. Adiagram of the Pye Unicam detector is shown in figure 41.

Nitrogen Nitrogen

Nitrogen

Coating Block

FromColumn

Evaporator Oven

FID

Detector Oven

Oxygen

Pyrolysis Oven

Hydrogen

Flow Restrictors

Cleaner Oven FeedSpool

Collection Spool

Figure 41 The Pye Unicam Moving Wire detector

The carrier wire from a spool passes through an oven at 750˚C and anyresidual lubricants remaining on the wire after drawing are burnt off.


68

After passing round a pulley the wire then goes through a coating blockwhere it is coated with a thin film of mobile phase. The wire then passesthrough an evaporator at about 105˚C (or a temperature appropriate forthe solvent(s) employed). In the evaporation chamber a nitrogen streamis allowed to flow counter-current to the wire movement to improve therate of solvent evaporation. The wire carrying the solvent-free solutethen enters (via a restriction) a pyrolysis tube that is maintained at about500˚C. The pyrolysis tube has a nitrogen supply entering at either endwhich sweeps the tube contents, including the pyrolysis products that areformed, out through a center tube into the FID.

The FID was a standard detector and the device functioned well and, asexpected, was completely independent of the nature of the solvents usedfor the mobile phase. The detector sensitivity was disappointing, andfound to be little better than the average refractive index detector viz. 5x 10-6 g/ml. The poor sensitivity resulted from excessive noise (not aweak signal) which may have been caused by high boiling impurities inthe solvents, fluctuations in the nitrogen flow and/or irregularities in thepyrolysis process. The linear dynamic range was also found to be lessthan two orders of magnitude. However, the device did establish theviability of the transport method for LC detection.


69

Synchronous Motor Gold Chain

Evaporation Tunnel

LC Column Coating Block

FIDAir Inlet

Gauze Jet

Figure 42. The Chain Detector

The Chain Detector

About the same time as James et al. developed the wire transportdetector, Haahti and Nikkari (34) described a similar device, simpler indesign, that employed a chain loop in place of the wire as a carrier. Adiagram of their apparatus is shown in figure 42. The carrier, a goldchain driven by a synchronous motor, passes through a coating blockwhere it is wetted with the column eluent. The chain then enters anevaporator tunnel, is heated, and the solvent volatilized leaving the solutedeposited on the chain. The chain then exits the tunnel into the actualflame of an FID.


70

EthylAlcohol

Surfactant

Mineral Oil

n-Heptane

Sample: mineral oil and a surfactant, solvent: n-heptane/ethyl alcohol, column: 2 x 300mm, column packing: silica gel, flow rate: 0.7 ml/min., chart speed: 24 cm/min.,evaporator temperature: 150˚C, nitrogen flow: 30 ml/min., hydrogen flow rate: 25ml/min., oxygen flow rate: 30 ml/min.

Figure 42. Chromatogram Obtained from the Chain Detector

Combustion takes place, ions are produced in the expected manner, andthe ion current processed in the usual way. Unfortunately, due to theocclusion of local, high concentrations of solute between the links of thechain, the detector output was extremely noisy and thus the systemexhibited a relatively poor sensitivity. A chromatogram obtained with thechain detector is shown in figure 42. The noise spikes on each peak areobvious, which, besides affecting the overall sensitivity of the detector,makes quantitative analysis extremely difficult.

The Modified Moving Wire Detector


71

In the early 1960s, a number of attempts were made to improve theperformance of the transport detector. In 1966 Karmen (35) introducedan aspirating system to draw the pyrolysis products into the hydrogenflame detector. Later in 1970, Scott and Lawrence (36) developed theKarman's system further and introduced a modified form of the originalmoving wire detector.

The sensitivity of the original detector, in addition to being degraded bythe high noise level, was also limited by the proportion of the pyrolysisproducts that entered the FID. Excepting synthetic polymers, (whichoften quantitatively produced monomers) many compounds yielded onlya few percent of volatile compounds on pyrolysis. Thus the FID couldonly sense a very small fraction of the products from the solutedeposited on the wire. If the solutes, were completely combusted inoxygen or air, however, then all the carbon in the solute would beconverted to carbon dioxide. Furthermore, if the carbon dioxide wasthen reduced to methane by mixing it with hydrogen and passing it overa nickel catalyst, the carbon dioxide would be quantitatively converted tomethane which could be detected by the FID. This procedure wouldincrease the sensitivity of the detector to those substances that gave pooryields on pyrolysis and, in addition, increase the linear dynamic rangeand possibly provide a predictable response.

A diagram of the original moving wire detector modified in this way isshown in figure 43. The hydrogen lines entering the detector body wereenlarged to reduce the flow impedance and permit the satisfactoryoperation of the aspirator. The detector was connected to a 2 in. lengthof 1/2 in I.D. thin walled stainless steel tube filled with about 2 g of nickelcatalyst and then to the aspirator. The wide tube was closed with a looseplug of quartz wool and .


72

OxygenCoating Block

FromColumn

Evaporator Oven

FID

Detector Oven

Air

Oxidation Oven

Hydrogen/Argon

Flow Restrictors

Cleaner Oven FeedSpool

Collection Spool

Molecular Entrainer

NickelCatalyst

Oxygen

Oxygen

Figure 43 The Modified Moving Wire Detector

The nickel catalyst was prepared by absorbing a saturated solution ofnickel nitrate onto 20/40 BS mesh brick dust, decomposing the nitrate at500˚C for 3 hours followed by reduction of the nickel oxide so producedto metal in a stream of hydrogen at 250˚C. The jet/venturi aspirator(supplied commercially as a molecular entrainer) was placed in line withthe hydrogen flow to the detector.

To improve the aspirating efficiency, the gas used was a mixture ofhydrogen and argon, and with this mixture, the jet/venturi pressure dropcontinuously sucked the combustion gasses into the hydrogen stream. Infigure 43 the reduced pressure side of the aspirator is shown connectedto the side limb of the oxidation tube and the two tube system used inthe original moving wire detector, was replaced by a single tube. Theoxygen or air is fed into the center of the tube and, thus, provided boththe evaporator flow and the oxidation flow. All tubes were constructedof quartz. The linear dynamic range of the system was determined to beabout four orders of magnitude a considerable improvement on the


73

pyrolysis instrument. The response index for a series of compounds ofdifferent chemical types ranged from 0.96 to 1.04. and the actualresponse was found to be proportional to the carbon content of thesolutes.

1 2 3 4 5

6

7 8 9 10 11 12Solvent

Injection

Figure 44 The Separation of Blood Liquids EmployingIncremental Gradient Elution and Monitored by the ModifiedMoving Wire Detector

However, due to the limited number of compounds that were tested thisrelationship should be assumed only with caution. A chromatogram ofblood lipids obtained by incremental gradient elution and monitored bythe modified detector is shown in figure 44. As incremental gradientelution involves a program of 12 solvents ranging from hydrocarbons,


74

chlorinated hydrocarbons, nitro-paraffins, esters, ketones and alcohols.This separation illustrates the versatility that is provided by this detectorfor solvent selection.

Van Dijk (37), attempted to improve the sensitivity of the detector byusing a spray procedure for coating the wire. The column eluent enteredan atomizer and spray from the nozzle was directed onto the wire. Thespray partly concentrated the solute by evaporation and also increase theload on the wire. A linear dynamic range of about 3 x 103 was claimedfor the system and a sensitivity increase of 50 over the original wiretransport detector (probably about 3 x 10-6 g/ml).

Yang et al. (38) also used a thermal spray for coating the wire and alsoclaimed an increased sensitivity. A heated chamber (through which theconduit from the column passed) was placed above a moving stainlesssteel belt. The solvent was rapidly brought to its boiling point resulting inspray leaving the exit of the conduit and coating the belt. Yang et al.also employed a photo-ionization detector and an electron capturedetector as alternatives to the FID.

Compton and Purdy (39) redesigned the Pye Unicam FID by inserting arubidium silicate glass bead above the flame and thus made its responsespecific and changed it into a nitrogen phosphorus detector. Stolyhwo etal. [40] employed metal spirals wound on wire and stranded wire toincrease the surface area of the carrier to increase the proportion of thecolumn eluent taken into the detector. A detectable mass of 100 ng oftriolei was claimed but the actual concentration sensitivity was notreported. Pretorious and Van Rensburg (41) tried to increase the carriertake-up by coating the wire with sodium silicate, kaolin and copperkaolin. some improvement to appears to have been realized. Slais andKrejei (42) replaced the normal FID with the NPD detector and used itto detect chlorine compounds. They mixed the combustion gases withhydrogen and passed the mixture directly into the NPD. At a columnflow rate of 0.37 ml/min., the sensitivity of the detector was stated to beabout 3 x 10-7 g/sec, which is equivalent, in concentration units, to about1.6 x 10-6 g/ml.


75

The moving wire detector has also been modified by Dugger (43) forradioactivity detection (e.g., detection of 14 carbon labeled compounds).To detect 14 carbon compounds, the solute on the wire was oxidized tocarbon dioxide and the radioactive gas passed to a Geiger-Muller tube.To detect tritium, the tritiated water produced on combustion was passedover heated iron to reduce it to hydrogen and tritium, which was thenalso passed to a Geiger-Muller tube.

More recently De Peña et al. (44, 45) have developed the transportdetector still further. A photograph of their detector, produced byScientific Detectors Ltd., is shown in figure 45.

Figure 45. The Uni-Mass Transport Detector

The major novel feature of this transport detector is the use of anoxidized titanium tape as the carrier which has a greatly improvedloading capacity and readily wets with both dispersive and highly polarsolvents (e.g., n-heptane and water). In addition, the coating, evaporating


76

and pyrolysis system has been made much more compact and the argondetector (which is used as the sensor) is made physically very close tothe pyrolyser (ca 2-3 cm) A photograph of the internal layout of thedetector is shown in figure 46.

Figure 46. The Internal Lay-Out of the Uni-Mass Detector

The performance of this detector remains to be established. It has foundgreat interest in biotechnology field as it responds to all substances andthe relative responses to different compounds appear to be within afactor of 5 and thus all components are detected.

This detector should have a theoretical limit of detection of about 1x10-9

g/ml (i.e. similar to the fluorescence detector) but this sensitivity has notbeen realized yet.

The Disc Detector

The disc detector was originally described by Dubsky (46) who used arotating gauze disc as the carrier. A diagram of the device is shown infigure 47. The rotating disc has a perimeter made of wire gauze and the


77

column exit is situated just above the gauze so that the eluent flowsthrough the gauze coating it with a film of mobile phase. The excess ofmobile phase is collected in a container situated below the disc. Aninfrared lamp, or some other appropriate heater, is placed a little aheadof the point of coating, in the direction of rotation of the disc whichevaporates the solvent leaving the solute coated on the gauze.

Gauze Disc

Column Heater

Electrodes

Flame FID

Synchronous Motor

Figure 47. The Disc Detector

The FID is placed diametrically opposite to the point of coating and theflame jet is situated beneath the gauze in such a manner that the flameitself is in contact with the gauze.

The FID electrodes are placed just above the gauze, directly over theflame. The ion current is collected by the electrodes and amplified in theusual way and fed to a recorder or data acquisition system. Thisarrangement is simple compared with the conventional wire transportdetector but although a tenfold increase in sensitivity is claimed, this isdifficult to confirm from the published data given. Szakusito andRobinson (47) reported that the metal gauze disc carrier was a source of


78

excessive noise. Specifically, significant "spikes" were caused by localconcentrations of solute accumulating at the intersections of the wiremesh during evaporation. These authors employed an alumina disc, 4.5in diameter, with the edge tapered to 0.25 mm thick. The thin edge wasused for coating and detection. It appeared that there a significantreduction in noise, but again, sensitivities were not given in terms thatwould allow comparison with other detectors. It appeared that incontinual use, the pores of the alumina could become blocked withresidue from incompletely combusted solutes or mobile phasecomponents and so the life of the alumina disc may well be limited. Thedisc does appear to be a simpler transport system than the wire or chainbut its reliability and sensitivity remain to be established.

The Evaporative Light Scattering Detector

Nebulizer Gas Eluent from Column

Nebulizer

Light Source

Fan

Exhaust

Light Trap

PhotomultiplierOptical Fibers

Courtesy of Polymer Laboratories Inc.Figure 48. A Diagram of the Evaporative Light ScatteringDetector


79

The evaporative light scattering detector, as its name implies, utilizes aspray that continuously atomizes the column eluent into small droplets.These droplets are allowed to evaporate, leaving the solutes as fineparticulate matter suspended in the atomizing gas.

The atomizing gas can be air or, if necessary, an inert gas. Thesuspended particles pass through a light beam and the scattered lightviewed at 45o to the incident light beam by means of a pair of opticalfibers. The scattered light transmitted through the fibers is sensed by aphotomultiplier and the output electronically processed and passed eitherto a computer data acquisition system or to a potentiometric recorder. Adiagram of the light scattering detector is shown in figure 48.

Eluent andAtomizing Gas

Atomized Stream of Particles

To Photocell Collimated Light Beam

Courtesy of Polymer Laboratories Inc.


80

Figure 49 The Sensor of a Commercial Evaporative LightScattering DetectorTheoretically, the detector responds to all solutes that are not volatile andas the light dispersion is largely Raleigh scattering, the response shouldbe proportional to the mass of solute present; as a consequence, it issometimes referred to as the mass detector. For a linear response, thedroplet size must be carefully controlled as it also determines the particlesize of the dried solutes. Detector sensitivity is claimed to be between 10and 20 ng of solute. However, in these terms it is difficult to comparewith other detectors. The great advantage of the detector, however, is itscatholic response and that its output is linearly related to the mass ofsolute present. However, the magnitude of the response does varywidely between different substances.

A diagram of the sensor of the evaporative light scattering detectormanufactured by Polymer Laboratories is shown in figure 49.

Courtesy of Polymer Laboratories Inc.

Peak Compound Mass (mg) Retent.Time(min).

1 cholesterol ester 5 0.7172 triglyceride 18 1.7463 cholesterol 10 4.6874 unknown 10 8.8605 phosphatidyl choline 10 10.0286 phosphatidyl -

ethanolamine17.390


81

Figure 50. The Separation of Some Lipid Class MaterialsMonitored by a Evaporative Light Scattering DetectorThe eluent is atomized in a stream of nitrogen and the finely dividedspray passes down a heated chamber during which time the solvent isevaporated. The removal of the solvent produces a stream of particleswhich then pass through a collimated beam of light. The scattered lightat an angle to the incident light is focused onto a photomultiplier tubeand the output is processed in an appropriate manner electronically. Thedevice is fairly compact and relatively simple to operate. An example ofthe results obtained when used for monitoring a general lipid classanalysis is shown in figure 50. The minimum detectable mass estimatedfrom this chromatogram appeared to be about 10 ng of solute. To someextent, this detector provides a replacement for the transport detector asit detects all substances irrespective of their optical or electricalproperties.

Liquid Light Scattering Detectors

Light scattering detectors differ from evaporative light scatteringdetectors in that they respond to the light scattered by a polymer orlarge molecular weight substance present in the column eluent itself. Thescattering is measured as it passes through an appropriate sensor cellwhile illuminated by a high intensity beam of light. The high intensitylight source is achieved by the use of a laser (light amplification by thestimulated emission of radiation) that also generates the light at theappropriate wavelength for measurement. There are two forms of thedetector: the low angle laser light scattering (LALLS) detector and themultiple angle laser light scattering (MALLS) detector. Both devicesare commonly used but the multiple angle laser light scattering detectoris more versatile as it provides molecular dimensions as well as themolecular weight of the eluted solute.

In the LALLS detector, the scattered light is measured at a very smallangle to the incident light (virtually 0o), and consequently the signal canbe affected by scattering from contaminating particulate matter present


82

as impurities in the eluent. This can cause significant noise, and thus,reduce the detector sensitivity. Discussions on the subject have beengiven by Wyatt (48) and some early experimental results reported by D.T. Phillips(49).

The ratio of the intensity of the light scattered at an angle (f), (If) to theintensity of the incident light (Io), for Rayleigh light scattering is given by

If

Io= a w Rf

where, (a) is the attenuation constant, (w) is a function of the refractive index,

and (Rf) is Rayleigh's constant

Thus, Rf =If

a w Io

Now the molecular weight (Mw) of the solute is related to the Rayleighfactor by the following expression,

Mw =Rf

c K - 2A 2Rf( )

where (c) is the concentration of the solute,(A2) is a function of polymer-polymer interactions,

and (K) is the polymer optical constant.

Substituting for (Rf)

Mw =

If

a w Io

c K - 2B2If

a w Io

Ê

Ë Á

ˆ

¯ ˜

=If

c a w Io K - 2B2 If( )

where

K =2 p2h2

l4 N dhdc

Ê Ë Á ˆ

¯ ˜

2

where (h) is the solvent refractive index,


83

(l) is the wavelength of the light in vacuum,and (N) is Avogadro's number.

Thus, a basic relationship can be developed between the molecularweight of the scattering material, the intensity of the scattered light andthe physical properties of the materials and equipment being involved.However, constants are in used, the magnitude of which are difficult todetermine. In practice a simple graphical procedure is used to determinethe molecular weight of the solute without the need to determine all thepertinent constants. Rearranging the equation

1Mw

=c K - 2A 2Rf( )

Rf=

cKRf

- 2cA2

or cKRf

= 2cA 2 +1

Mw

Now (c), (K), and (Rf) are either known or can all be calculated from

known data and light scattering measurements; thus, by plotting cKRf

Ê

Ë Á

ˆ

¯ ˜

against (c) a straight line will be produced with the intercept being1

Mw

Ê

Ë Á

ˆ

¯ ˜ .

The Low Angle Laser Light Scattering Detector

The optical system of the low angle laser light scattering detectorproduced by LDC Analytical of the Thermo Instruments Corporation isshown diagramatically in figure 51. To conserve space, a folding prism isused that allows the device to be contained to a reasonable size yetaccommodate the length of the laser generator. Light from the laserpasses through a diverging lens, through a chopper and then through thefolding prism. On leaving the prism the beam passes through somemeasuring attenuators and a calibrating attenuator shutter and thenthrough the cell. An annular mask is situated between the cell and therelay lens and only allows light scattered in the cell at a low angle to pass


84

to the relay lens. Between the annular shutter and the relay lens is asafety attenuator that ensures that none of the potentially damaging laserlight can reach the photomultiplier. The scattered light is focused througha field stop onto the forward detector lens. Between the field stop andthe forward detector lens is a prism that allows the scattered light to beviewed through a microscope.

Diverging LensChopper

Folding Prism

MeasuringAttenuators Annulus

Mask

CalibratingAttenuator/ Shutter

CellWindows

Condensing Lens

Apertures

SafetyAttenuator

RelayLens

FieldStop

ViewingMicroscope

ForwardDetector Lens

RearDetector Lens

SensorAperture

Photomultiplier

FilterHolder

AnalyzerPolarizer

Opal Diffuser

RedFilter

LASER

Courtesy of LDC Analytical, Thermo Instruments Corporation.

Figure 51 Optical Diagram of a Low Angle Laser Light ScatteringDetector

Between the forward detector lens and the rear detector lens is A filterholder and an analyzer/polarizer. Finally the light is focused through asensor aperture to an opal diffuser that spreads the scattered lightthrough a red filter and onto the photomultiplier.


85

The device is frequently used with a refractive index detector in series tocoincidentally measure the refractive index of the eluent. This isnecessary to calculate (K) from the refractive index. As discussed, themolecular weight of a solute is determined from the intercept of thegraph relating ( cK

Rf) to the solute concentration (c) as shown in figure

52. The concentration calculated from the refractive index detectoremploying the data from prior calibration.

1M w

cKRf

c

cK2A 2 Rf

Figure 52 Determination of Molecular Weight from Low AngleLight Scattering Measurements

The detector sensitivity appears to be similar to that of the refractiveindex detector and with about the same linearity. However, the greatestadvantage of this detector is that it can provide molecular weight datafor extremely large molecules.

The Multiple Angle Laser Light Scattering (MALLS) Detector

The multiple angle laser light scattering detector differs from the lowangle device, in that scattering measurements are made at a number ofdifferent angles, none of which are close to the incident light. Thissignificantly reduces problems associated with light scattering from


86

particulate contaminants. Data taken at different angles to the incidentlight allows the root-mean-square (rms) of the molecular radius r2 1/ 2

to be calculated in addition to the molecular weight of the substance. Therelationship that is used is as follows:

cKRf

= a r2 1/ 2sin q( )2 + b Mw

Theory can provide explicit functions for (a) and (b) but these constantsare obtained in practice by the use of calibrating standards of knownmolecular weights and molecular radii. In addition, a photocell will nothave precisely the same response to low light intensities and calibrationprocedures are also necessary to account for their different responses.

The number of different angles of measurement differs with differentinstruments, and some measure the scattered light intensity at 16different angles. In general, the more data points taken at differentangles, the more precise the results will be. A diagram of a (MALLS)detector system which measures the light scattered at three differentangles is shown in figure 53.

LASER

Glass Cell

LASER Beam

Sensor 2

Sensor 3Sensor 1


87

Courtesy of Wyatt Technology Corporation

Figure 53 The Multiple Angle Laser Light Scattering Detector(miniDawn®)

This device (the miniDawn®), manufactured by Wyatt TechnologyCorporation, contains no mirrors, prisms or moving parts and the lightpaths are direct and not "folded". Light passes from the laser(wavelength 690 nm) directly through the sensor cell. Light scatteredfrom the center of the cell passes through three narrow channels to threedifferent photocells, set at 45o and 90o and 135o to the incident light.Thus scattered light is continuously sampled at three different anglesduring the passage of the solute through the cell.

A continuous analog output is provided from the 90o sensor and all thesensors are sampled every 2 sec. The molecular weight range extendsfrom 10 3 to 10 6 Daltons and the rms radii from 10 to 50 nm. Thetotal cell volume is about 3 ml and the scattering volume is 0.02 ml. Thedetector has a sensitivity, defined in terms of the minimum detectableexcess Rayleigh ratio of 5 x 10-8 cm-1. This is difficult to translate intonormal concentration units but appears to be equivalent to a minimumdetectable concentration of about 10-6 g/ml.

KcR(q )

0 0.5 1.0sin ( q/2)2

Intercept gives a Value for the Molecular Weight

Slope Gives a Value for the rms Molecular Radius

Figure 54. Calibration Curves


88

The relationship between the intensity of the scattered light, thescattering angle and the molecular properties are as follows:

cKRf

= 2cA 2 +1

Mw P(f) where P(q) describes the

dependence of the scattered light on the angle of scatter and the othersymbols have the meanings previously attributed to them.The relationship between the angle of scattering, (q), the molecularweight and the rms molecular radius of the solute is obtained using theappropriate equation. Employing suitable reference materials, graphs ofthe form shown in figure 54 can be constructed to evaluate constants (a)and (b) and thus permit the measurement of the molecular weight andmolecular radius of unknown substances.

The Electrical Conductivity Detector

The electrical conductivity detector measures the conductivity of themobile phase. There is usually background conductivity which must bebacked-off by suitable electronic adjustments. If the mobile phasecontains buffers, the detector gives a base signal that completelyoverwhelms that from any solute usually making detection impossible.Thus, the electrical conductivity detector a bulk property detector. andsenses all ions whether they are from a solute or from the mobile phase.

In order to prevent polarization of the sensing electrodes, AC voltagesmust be used and so it is the impedance not the resistance of theelectrode system that is actually measured. From a physical chemistrystand point the conductivity of a solution is more important than itsresistance. However, it is the resistance (impedance) of the electrodesystem that determines the current across it.

The resistance (R) of any conductor varies directly as its length (L) andinversely as its cross sectional area (a).

Thus, R =r La

where (r) specific resistance of the conductor.


89

The specific resistance of a conductor is the resistance across twoopposite faces of a 1 cm cube made of the conductor material. Thespecific conductance (k) of a solute is the reciprocal of the specificresistance, i.e., k =

1r

Thus the conductance of a given solution (C) is given by,

C =1R

=k aL

ohms -1

The conduction of all the ions produced by 1 g equivalent of anelectrolyte at any particular concentration can be evaluated by imaginingtwo large parallel electrodes, 1 cm apart, and the whole of the solutionplaced between them. The conductivity of the system is called theequivalent conductance (L).

If 1 g equivalent of the electrolyte is dissolved in (v) ml of solution; thenit follows that this will cover (v) cm2 of electrode area. It follows that inthe above equation (a) becomes (v) and (l) is unity and

L =1000k

cohms-1cm2

where (c) is the electrolyte concentration in gram equivalents /liter.

or k =Lc

1000

Thus, C =

1R

=L c a

1000 Lohms-1

Thus, the cell resistance is inversely proportional to the electrolyteconcentration (c), the equivalent conductance of the electrolyte and thecell geometry [(a) and (L)]. Thus the cell output (which is, in factproportional to the resistance change of the cell) must be modified toprovide an output linearly related to concentration.

The basic sensor, being so simple, has changed little from that firstdescribed (1). The sensor volume has been reduced but still consists of


90

two electrodes situated in the column eluent, the resistance (or strictly theimpedance) between which is measured by an appropriate electroniccircuit. Figure 55 shows three simple forms of the conductivity sensoralthough many other geometric forms have been described in theliterature. The basic sensor is, indeed, very simple, the top design is easilyconstructed and can have an effective sensing volume of only a fewmicroliters.

Electrodes FromColumn To Waste

FromColumn

To Waste

PTFE Insulating Tube

PTFE Waste Tube

Stainless SteelTube. (Electrode Grounded)

Stainless SteelTube. (Electrode to Amplifier)

FromColumn To

Waste

PTFE Insulating Tube

PTFE Waste Tube

Stainless SteelTubes. (Isolated Electrodes)

SensorVolume

SensorVolume

Figure 55. The Electrical Conductivity Detector

The center sensors are made from short lengths of stainless steel tubing(about 0.020 in I.D. and 1/16 in O.D). The two tubes slide into a PTFE


91

tube and the space (inside the PTFE tube) between the two tubes issensing volume. The lower sensor is very similar, but two lengths ofstainless steel tube are employed, both of which are isolated from anyother metal connection by PTFE tubing. The two electrodes are, thus,electrically isolated (this can be advanatgeous for certain types ofelectronic measuring circuits). These sensors have sensing volumes ofonly a few nanoliters and are, thus, often used in detectors for capillaryelectrophoresis.

The electrical system consists of a frequency generator (1,000-5,000 Hz)that provides an AC potential across the cell. The sensor is placed in onearm of a Wheatstone bridge as shown in figure 56.

Sensing Cell

Out-of-Balance Signal

AC Supply

If a single sensor is employed, it is situated in one arm of the bridge; if, however, areference sensor is also used, then the two sensors are situated in opposite arms of thebridge. In either case the out–of–balance signal can be passed to a precision rectifierand the output either handled by an analog circuit or sampled directly by thecomputer. Sometimes a variable resistance is situated in one of the other arms of thebridge and is used for zero adjustment to back-off any signal arising from ionscontained by the mobile phase.

Figure 56 One Form of the Wheatstone Bridge Used with theElectrical Conductivity Detector


92

The out-of-balance signal is rectified and the DC signal either passed to anonlinear amplifier or a computer data acquisition system. The non-linear amplifier is necessary to render the output linearly related to ionconcentration. If the output is passed directly to the computer, then thelinearization can be carried out with appropriate software and thechromatogram is presented on the printer.An example of a separation monitored with a conductivity detectoruseing an ion supression technique (see book 15) is shown in figure 57.

Time 10 Minutes

Courtesy of Dionex Inc.

A proprietary ion exchange column, IonPacCS12, was used and the mobile phaseconsisted of a 20 nM methanesulphonic acid solution in water. A flow rate of 1ml/min was employed and the sample volume was 25 ml.

1. Lithium 2. Sodium 3. Ammonium4. Potassium 5. Magnesium 6. Calcium

Figure 57. Determination of Alkali and Alkaline Earth Cations

The electrical conductivity detector has a sensitivity (minimum detectableconcentration) of about 5 x 10-9 g/ml and a linear dynamic range ofabout 200 where 0.97 < r < 1.03. It is used in probably over 95% of all


93

analyses involving ion exchange procedures to separate inorganic andorganic ions.

The Electrochemical Detector

The electrochemical detector responds to substances that are eitheroxidizable or reducible and the electrical output is an electron flowgenerated by a reaction that takes place at the surface of the electrodes. Ifthe reaction proceeds to completion (exhausting all the reactant) thecurrent becomes zero and the total charge generated will be proportionalto the total mass of material that has been reacted. This process is calledcoulometric detection. If, however, the mobile phase is flowing past theelectrodes, the reacting solute will be continuously replaced as the peakpasses through the detector. All the time there is solute present betweenthe electrodes, a current will be maintained, albeit varying in magnitude.Until relatively recently, this procedure was that most commonemployed in electrochemical detection and is called amperometricdetection.

The electrochemical detector requires three electrodes, the workingelectrode (where the oxidation or reduction takes place), the auxiliaryelectrode and the reference electrode (which compensates for anychanges in the background conductivity of the mobile phase). Theprocesses taking place at the electrode surface can be very complex;nevertheless, the dominant reaction can be broadly described as follows.At the actual electrode surface the reaction is extremely rapid andproceeds almost to completion. This results in the layer close to theelectrode being virtually depleted of reactant. As a consequence, aconcentration gradient is established between the electrode surface andthe bulk of the solution. This concentration gradient causes solute todiffuse into the depleted zone at a rate proportional to the soluteconcentration in the bulk of the mobile phase. Thus, the currentgenerated at the electrode surface will be determined by the rate atwhich the solute reaches the electrode and consequently, as the processis diffusion controlled, will depend on solute concentration and the


94

magnitude of solute diffusivity. The response of the detector or current(i) is described by the following equation,

i = n F A K T c ua where (n) is the number of electrons per molecule involved in the

reaction, (F) is the Faraday Constant,(A) is the area of the working electrode,

(KT) is the limiting Mass Transfer Coefficient,(c) is the solute concentration,(u) is the linear velocity of the mobile phase over the surface

of the electrode,and (a) is a constant usually taking a value between 1/3 and 1/2.

It is seen that the current (i) (and, thus, the sensitivity), can be raised byeither increasing the electrode area, increasing the transfer coefficient orincreasing the velocity of the mobile phase past the electrodes. It wouldappear that increasing the electrode area would be the easiest, however,increasing the electrode surface area while maintaining an amperometricresponse also increases the noise, often to such an extent that there is anoverall reduction in detector sensitivity. Weber and Purdy (50) andHanekamp et al. (51) showed that under certain conditions a reductionin the sensor size produces a significant increase in signal-to-noise and,thus an increase in sensitivity. Higher flow rates will also increase the rateof solute transfer and the sensitivity which would be an added advantageto miniaturization. However, the sensor will be very flow sensitive andthus the flow rate must be kept very constant and the detector wouldnot be amenable to flow programming development. Miniaturizationwould also reduce peak dispersion in the sensor.

Electrode Configurations

The electrodes can take a number of different geometric forms whichhave been described in some detail by Poppe (52). Some examples ofdifferent electrode configurations are shown in figure 57. Examples (A)and (B) are two common forms of thin-layer cells. (A) has the working


95

electrode sealed into the cell wall with the reference and auxiliaryelectrodes situated down-stream to the working electrode. (B) is similarto (A) but with the auxiliary electrode sealed in the wall of the tubeopposite the working electrode and, again, the reference electrode down-stream. (C) is a wall-jet electrode where the eluent is allowed to impingedirectly onto the working electrode which is situated opposite the jet.This arrangement not only increases the value of (u) (the velocity of theliquid passing over the electrode and thus the transfer coefficient (KT))but also scrubbs the surface of the working electrode helping to reducethe need for frequent cleaning. (D) and (E) are two examples ofcylindrical electrodes; in (D) the working electrode is in the form of a rodstretching across the diameter of the sensor cell and in (E) the workingelectrode comprises an annular ring set in the cell wall.


96

Working Electrode

Auxilary and ReferenceElectrodes Downstream

Reference ElectrodesDownstream

AuxilaryElectrode

Working Electrode

A B

Auxilary and Reference ElectrodesDownstream

WorkingElectrode E

Working Electrode

Auxilary and Reference ElectrodesDownstream

Flow from Column Working

ElectrodeAuxilary and Reference ElectrodesDownstream

C D

Figure 57. Different Electrode Configurations

In both cases the auxiliary and the reference electrodes are situateddown-stream to the working electrode.

Electrode Construction

The choice of electrode construction material is restricted due to theneed for mechanically ruggeness and long term stability. The mostcommon material is carbon paste made from a mixture of graphite andsome suitable dielectric substance. This material has the disadvantage thatit is soluble in some solvents, although, using special waxes or polymersas dielectric binders to contain the graphite, helps reduce the solubilityproblem. Vitreous or 'glassy' carbon is an excellent electrode materialparticularly if organic solvents are to be used and is probably the most


97

popular contemporary electrode material. Glassy carbon is produced byslowly baking a suitable resin at elevated temperatures until it iscarbonized and then heating it to a very high temperature to causevitriation. Vitreous carbon is relatively pure, mechanically strong, hasgood electrical properties and can be readily cleaned mechanically. It alsoperforms particularly well when operated at a negative potential. Glassycarbon electrodes are preferable to carbon paste electrodes due to theirinherent resistance to solvents.

Electrochemical detection imposes certain restrictions on the type ofchromatography that is employed and the mobile phase that is used. Thedetecting system requires a conducting mobile phase and thus mustcontain water. Thus, the majority of 'normal phase' systems are notusable. In addtion, very high solvent concentrations may render itinsufficiently conducting. Reversed phase chromatography, however, isideally suited to electrochemical detection. Nevertheless, certainprecautions must be taken for its effective use. The mobile phase must becompletely oxygen free which can be removed by bubbling heliumthrough the solvent reservoir. It is also important to remove oxygenfrom the sample before making an injection. The solvents must also befree of metal ions or serious base line instability will result. Non-aqueoussolvents such as pure acetonitrile can be employed but certain salts liketetrabutyl-ammoniumhexafluorophosphate must be added to render thesolvent conducting.Basic Electrochemical Detector Electronics

A simplified form of the circuit that is in general use is shown in figure58. The amplifiers can best be considered as FET operational amplifierssuch as the TLO81. The auxiliary electrode is held at a fixed potential bythe first amplifier, the voltage being selected by the potentiometer (P)that is connected to a regulated power supply. The current flowingthrough the working electrode is processed by the second amplifier andthe output fed to the recorder or data acquisition system. Theelectrochemical detector as described is extremely sensitive but hascertain disadvantages.


98

A

P

Regulated Power Supply

WR

Amp.2

Amp.1

Output

Feed BackResistance

Figure 58 Basic Circuit Used with Electrochemical Detectors

The mobile phase must be extremely pure, oxygen free and devoid ofmetal ions. A more serious problem arises from the adsorption of theoxidation or reduction products on the working electrode surface. As aresult, the electrode system must be frequently calibrated to ensureaccurate quantitative analysis. In addition, the detector must be regularlydissembled and cleaned (usually by a mechanical abrasion). Manyattempts have been made to avoid this contamination problem but,although it has been reduced, it has not been completely eliminated(particularly in the amperometric form of operation). Due to potentiallylow sensing volume the detector is very suitable for use with small borecolumns. An example of the use of the electrochemical detector tomonitor the separation of a series of catacholamines is shown in figure59. It is seen that the detector is very sensitive but its value in g/ml(which would allow some comparison with other detecting methods) cannot be evaluated from the data.


99

1

0

2

3

4

5

6

7

8 9

1011

12

13

14

15

16

17 18

5 15

6

20Time (Minutes)

100 pA


Column: HC-3 C18 (100 mm x 4.6 mm); mobile phase: aqueous solution of 100 nMformic acid , 0.35 nM octane sulphonic acid, 1.0 nM citric acid 0.10 nM EDTA, 5%acetonitrile, 0.25% v/v diethylamine, pH to 3.10 with potassium hydroxide; flow rate 1ml/min; detection: oxidative amperometric with glassy carbon electrode at 100 mVpotential vs. Ag/AgCl electrode.

1. 3,4 dihydroxymendelic acid 200 pg 10. dopamine 200 pg2. L-dopa 600 pg 11. metanephrine 400 pg3. vanillymendelic acid 400 pg 12.3,4-dihydroxyphenylacetic acid

200 pg4. norepinephrin 200 pg 13. N-methyl dopamine 400 pg5. a-methyl dopa 600 pg 14. tyramine 1 ng6. 3-methoxy,4-hydroxyphenylglycol 400 pg

15. 5-hydroxyindole-3-acetic acid 200 pg

7. epinephrine 200 pg 16. 3-methoxytyramine 400 pg8. 3,4-dihydroxybenzylamine 200 pg 17. 5-hydroxytryptamine 200 pg9. normetanephrin 400 pg 18. homovanillic acid 400 pg


100

Figure 59. The Separation of Some Catacholamines Monitored byan Electrochemical Detector.

The column appears to be packed with 3 micron particles which shouldprovide an efficiency of ca. 16,000 theoretical plates. This high efficiencymeans the peaks would have a very small volume, thus, even with thesmall peak masses, the peak concentrations could be relatively high.Quoting sensitivities in terms of the mass contained in each peak willalways give an enhanced impression of detector sensitivity when a highefficiency column is used.

The Multi–Electrode Array Detector

The success of the electrode array as an LC detector is probably due tothe development of the porous carbon electrode. This electrode is madeof porous graphitic carbon, which has a very high surface area, ismechanically robust and, more important, is permeable to the mobilephase. As a consequence, flow through electrodes can be constructed.The material ideal for electrochemical detection in a number of ways. Asthe surface area is greatly in excess of that required for efficientelectrochemical reaction, it can be severely contaminated before it fails tofunction. In fact, as much as 95% of the surface can be contaminatedbefore it requires cleaning. When the electrode becomes sufficientlycontaminated to require cleaning (which, according to the manufacturers,may occur between one and three years of continual use), thecontamination can be rapidly removed by flushing with nitric acid.

The porous graphitic carbon electrode facilitates the construction ofelectrode arrays. In use, the large surface area of the porous electrodeensures that 100% of the eluted material is reacted. Thus, theelectrochemical reaction is no longer amperometric, but nowcoulometric. This is an important difference and makes the array systempractical. The electrode system is shown diagramatically in figure 60.Each electrode unit has a central porous carbon electrode, on either sideof which is situated a reference electrode and a auxiliary electrode. Asthe pressure drop across the porous electrode is relatively small, theseelectrode units can be connected in series forming an array. Up to 16


101

units can be placed in series and these arrays are commercially available.However, a sensor system that contains as many as 80 electrodes in thespace of a few millimeters has also been constructed (53).

ReferenceElectrodesReference

Electrodes

CounterElectrodes

CounterElectrodes

Porous Working Electrodes

Courtesy ESA Inc.

Figure 60. The Coulometric Electrode System Employing PorousGraphitic Carbon Electrodes

A progressively greater potential is applied sequentially to the electrodesof each consecutive unit. This results in all the solutes migrating throughthe array until each reaches the unit that has the required potential topermit its oxidation or reduction.

A comparison between amperometric and coulometric detection isshown in figure 61. In amperometric detection only a small part of thesolute is reacted and so the remainder can proceed to the next electrodesystem and be detected again. In this way each of the units will detect allthe solutes and the graph shown in the upper part of the figure isproduced. It is seen that there is no discrimination by the different


102

electrode voltages. Coulometric detection results in the total solute beingreacted and thus it will be detected by that unit that has the requiredpotential and not be sensed by other units.

Electrode Voltage

Electrode VoltageRetention Time

Retention Time

Courtesy of the Analyst.

Figure 61 Three Dimensional Graphs Demonstrating theDifference between Amperometric and Coulometric DetectionEmploying an Electrode Array

It follows that distinct peaks will be produced only at those units with theappropriate potentials. The result is a three dimensional graph similar tothat shown in the lower part of the figure. It is seen that each unitproduces a peak at a potential that is characteristic of the solute beingdetected. Although the voltage increases progressively from one unit tothe next, reaction is not completed at one unit only. This is becausereaction will not take place at a specific potential, but over a narrow


103

range of potentials. In practice the signal is usually detected over threecontiguous electrode units. For example, the first will oxidize a verysmall amount of the solute, the second unit will be the dominant unit andoxidize the majority of the solute, while the third unit will oxidize thesmall remaining quantity of solute.

1 2 3 4 5 6 7 8

1 2 3 4 5 6 7 8

Electrode Units

Electrode Units

Courtesy of the Analyst.

Figure 62. Use of the Electrochemical Array to ConfirmCompound IdentityThis produces a characteristic pattern of peaks for a particular solute.The ratio of peak height for the three contiguous electrode units thatsense the substance will be different for different substances althoughthey may be reacted at the same three electrodes. This is obviously a


104

method of confirming the identity of the solute and is demonstrated infigure 62. The upper graph shows the reference chromatogram of twostandards each showing specific retention times and a specific peakpattern as provided by the electrode array detector. The lower graph isfor a similar sample and, although the retention times for the pair ofsolutes is very similar, the pattern given by the electrode array detectorclearly shows that the second compound eluted is not the same as that ofthe second standard.

2Retention Time (minutes)

0 10 20

203 1

5 422

1319

1815

2730

29

17 6 7

2326

12 9

1025

14

11 8

16

21 2428

Mobile Phase 1% Methanol to 40% Methanol in a Phosphate (0.1 mol l-1 buffer withion pairing (pH 3.4)1. Dihydroxyphenylacetic acid 16. Metenephrine2. Dihydroxyphenylethylene glycol 17. Methoxyhydroxyphenyl glycol3. L-Dopa 18. Methoxytyramine4. Dopamine 19. N-methylserotonin5. Epinephrine 20. Norepinephrim6. Guanine 21. Normetenephrine7. Guanosine 22. Salsolinol8. Homovanallic acid 23. Octopamine9. Hydroxybenzoic acid 24. Seratomin10. Hydroxyindoleacetic acid 25. Tryptophan11. Hydroxyphenylacetic acid 26. Tyrosine12. Hydroxyphenyllactic acid 27. Uric Acid13. Hydroxytryptophan 28. Vanillic acid


105

14. Kynurenine 29. Vanylmandelic acid15. Melatonin 30. Xanthine

Courtesy of the Analyst.Figure 11 The Separation of 30 Neuroactive Substances Monitoredby an Electrochemical Array

The electrode array detector also gives improved apparentchromatographic resolution in a similar way to that of the diode arraydetector. Two peaks that have not been chromatographically resolvedand are eluted together can still be shown as two peaks that are resolvedelectrochemically and can be quantitatively estimated. Anotheradvantage is that high oxidation potentials can be used without the highbackground currents and noise that usually accompany such operatingconditions. The electrodes that are operating at high voltages are"buffered" by the previous electrodes operating at lower voltages whichresults in reduced background currents and noise.

Another example of the application of the detector to the separation of anumber of neuroactive substances (54)is shown in figure 63. It is seenthat for certain applications the electrochemical array detector can beextremely useful.Nevertheless, in order to use the detector, the solutesmust be amenable to electrochemical reaction and capable of beingseparated using a mobile phase that will conduct an ion current

References 1. A. Tiselius and D. Claesson, Arkiv. Kemi Minearl. Geol., 15B(No. 18)(1942) 2. A, J. P. Martin and S. S. Randall, Biochem. J., 49(1951)293 4. M. J. E. Golay, "Gas Chromatography 1958", (Ed. D. H. Desty), Butterworths, London (1958)36. 5. J. G. Atwood and M. J. E. Golay, J. Chromatogr., 218(1981)97. 6. R. P. W. Scott and P. Kucera, J. Chromatogr. Sci.,18(1971)641. 7. C. H. Lochmuller and M. Sumner, J. Chromatogr. Sci.,18(1980)159. 8. I. Halasz, H. O. Gerlach, K. F. Gutlich and P. Walkling, US Patent


106

3,820,660, (1974). 9. R. Tijssen, Separ, Sci. Technol., 13(1978)681.10. E. D. Katz and R. P. W. Scott, J. Chromatogr., 268(1983)169.11. J. G. Atwood and M. J. E. Golay, J. Chromatogr., 218(1981)97.12. R. P. W. Scott, P Kucera and M. Munroe, J. Chromatogr., 186(1979)475.13. A. Tiselius and D. Claesson, Ark. Kemi. Mineral. Geol. 15B(No.18)(1942).14. C. Christiansen, Ann. Phys. Chem., 3(1884)298.15. C. Christiansen, Handbook of Chemical Microscopy (Ed. E. M. Chanot and C. W. Mason), John Wiley and Sons, New York, Vol. 1((1958)101 and 189.16. M. Bakken and V. J. Stenberg, J. Chromatogr. Sci., 9(1971)603.17. 7. J. P. Gorden, R. C. C. Leite, R. S. Moore, S. P. S. Posto, J. R. Whinnery, Bull. Am. Phys. Soc., (2) 9(1964)501.18. J. P. Gorden, R. C. C. Leite, R. S. Moore, S. P. S. Posto, J. R. Whinnery, J. Appl. Phys., 36(1965)3.19. C. E. Buffet and M. D. Momis, Anal. Chem. 54(1982)1824.20. R. A. Grant, J. Appl. Chem. 8(1959)136.21. M. Poppe and J. Kunysten, J. Chromatogr. Sci. 10(1972)16A.22. L. V. Benningfield Jr., Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy, March 5-9(1979)paper 123.23. L. V. Benningfield Jr. and R. A. Mowery Jr., J. Chromatogr. Sci., 19(1981)115.24. 14. R. K. Bade, L. V. Benningfield Jr., R. A. Mowery and E. N. Fuller, Am. Lab.13(10)(19810130.25. J. P. Gorden, R. C. C. Leite, R. S. Moore, S. P. S. Posto, J. R. Whinnery, Bull. Am. Phys. Soc., (2) 9(1964)501.26. J. P. Gorden, R. C. C. Leite, R. S. Moore, S. P. S. Posto, J. R. Whinnery, J. Appl. Phys., 36(1965)3.


107

27. C. E. Buffet and M. D. Momis, Anal. Chem. 54(1982)1824.28. M. Poppe and J. Kunysten, J. Chromatogr. Sci. 10(1972)16A.29. L. V. Benningfield Jr., Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy,March 5-9(1979)p 123.30. L. V. Benningfield Jr. and R. A. Mowery Jr., J. Chromatogr. Sci., 19(1981)115.31. R. K. Bade, L. V. Benningfield Jr., R. A. Mowery and E. N. Fuller, Am. Lab.13(10)(19810130.32. J. M. Miller and R. Strusz, Am. Lab. Jan(1970)2933. A. T. James, J. R. Ravenhill and R. P. W. Scott, Chem. Ind. (1964)746.34. E. O. A. Haahti and T. Nikkari, Acta. Chem. Scand., 17(1963)2565.35. A. Karmen, Anal. Chem. 38(1966)286.36. R. P. W. Scott and J. F. Lawrence, J. Chromatogr. Sci.,8(1970)65.37.12. 14. L. M. Van Dijk, J. Chromatogr. Sci. 10(1972)31.38. L. Yang, G. J. Ferguson and M. L. Vertal, Anal. Chem.,56(1984)2632.39. B. J. Compton and W. C. Purdy, J. Chromatogr., 169(1979)39.40. A. Stolyhwo, O. S. Privet and W. L. Erdahl, J. Chromatogr. Sci. 8(1970)65.41 V. Pretorius and J. F. van Rensburg, J Chromatogr. Sci., 11(1973)263.42. K. Slais and M Krejci, J. Chromatogr., 91(1974)181.43. N. A. Dugger, US Patent 528,343.44. De Peña45 De Peña46. H. Dubsky, J. Chromatogr.,71(1972)395.47. J. J. Szakusito and R. E. Robinson, Anal. Chem., 46(1974)1648.48. P. J. Wyatt, Applied Optics 7(196801879.


108

49. D. T. Phillips, Nature 221(1969)1257.50. S. G. Weber and W. C. Purdy, Ind. Eng. Chem. Prod. Res. Dev., 20(1981).51. H. G. Hanekamp, P. Boss and R. W. Frei, Trends in Anal. Chem, 1(1982)135.52. H. Poppe, Instrumentation for High-Performance Liquid Chromatography, (Ed. J. F. K. Huber) Elsevier, Amsterdam and New York (1978).53. A. Atsushi, T. Matsue and I. Uchida, Anal Chem., 64 (1992)44.54. C. N. Svendsen, Analyst, 118(Feb)(1993)123.

LC Detector

Documents

perkin elmer

major emission

bulk property

high speed

linear dynamic

high flow

reversed phase

minimum detectable